Upload folder using huggingface_hub
Browse files- README.md +37 -0
- checkpoint-456/config.json +47 -0
- checkpoint-456/model.safetensors +3 -0
- checkpoint-456/optimizer.pt +3 -0
- checkpoint-456/rng_state.pth +3 -0
- checkpoint-456/scheduler.pt +3 -0
- checkpoint-456/trainer_state.json +3450 -0
- checkpoint-456/training_args.bin +3 -0
- config.json +47 -0
- model.safetensors +3 -0
- runs/Jun30_09-14-46_r-bastien2-gt-inno-2d039o9i-771b3-8acwj/events.out.tfevents.1751274888.r-bastien2-gt-inno-2d039o9i-771b3-8acwj.108.0 +2 -2
- runs/Jun30_09-14-46_r-bastien2-gt-inno-2d039o9i-771b3-8acwj/events.out.tfevents.1751302564.r-bastien2-gt-inno-2d039o9i-771b3-8acwj.108.1 +3 -0
- special_tokens_map.json +15 -0
- spiece.model +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +58 -0
- training_args.bin +3 -0
- training_params.json +30 -0
README.md
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
library_name: transformers
|
| 4 |
+
tags:
|
| 5 |
+
- autotrain
|
| 6 |
+
- text-classification
|
| 7 |
+
base_model: albert/albert-base-v2
|
| 8 |
+
widget:
|
| 9 |
+
- text: "I love AutoTrain"
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Trained Using AutoTrain
|
| 13 |
+
|
| 14 |
+
- Problem type: Text Classification
|
| 15 |
+
|
| 16 |
+
## Validation Metrics
|
| 17 |
+
loss: 0.7394976019859314
|
| 18 |
+
|
| 19 |
+
f1_macro: 0.7635683760683761
|
| 20 |
+
|
| 21 |
+
f1_micro: 0.7666666666666667
|
| 22 |
+
|
| 23 |
+
f1_weighted: 0.763568376068376
|
| 24 |
+
|
| 25 |
+
precision_macro: 0.7832887700534759
|
| 26 |
+
|
| 27 |
+
precision_micro: 0.7666666666666667
|
| 28 |
+
|
| 29 |
+
precision_weighted: 0.783288770053476
|
| 30 |
+
|
| 31 |
+
recall_macro: 0.7666666666666666
|
| 32 |
+
|
| 33 |
+
recall_micro: 0.7666666666666667
|
| 34 |
+
|
| 35 |
+
recall_weighted: 0.7666666666666667
|
| 36 |
+
|
| 37 |
+
accuracy: 0.7666666666666667
|
checkpoint-456/config.json
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "albert/albert-base-v2",
|
| 3 |
+
"_num_labels": 4,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"AlbertForSequenceClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_probs_dropout_prob": 0,
|
| 8 |
+
"bos_token_id": 2,
|
| 9 |
+
"classifier_dropout_prob": 0.1,
|
| 10 |
+
"down_scale_factor": 1,
|
| 11 |
+
"embedding_size": 128,
|
| 12 |
+
"eos_token_id": 3,
|
| 13 |
+
"gap_size": 0,
|
| 14 |
+
"hidden_act": "gelu_new",
|
| 15 |
+
"hidden_dropout_prob": 0,
|
| 16 |
+
"hidden_size": 768,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "CLE_DSI_SIEP",
|
| 19 |
+
"1": "CLE_DSI_SIL",
|
| 20 |
+
"2": "CLE_DSI_SPIA",
|
| 21 |
+
"3": "CLE_DSI_SSUR"
|
| 22 |
+
},
|
| 23 |
+
"initializer_range": 0.02,
|
| 24 |
+
"inner_group_num": 1,
|
| 25 |
+
"intermediate_size": 3072,
|
| 26 |
+
"label2id": {
|
| 27 |
+
"CLE_DSI_SIEP": 0,
|
| 28 |
+
"CLE_DSI_SIL": 1,
|
| 29 |
+
"CLE_DSI_SPIA": 2,
|
| 30 |
+
"CLE_DSI_SSUR": 3
|
| 31 |
+
},
|
| 32 |
+
"layer_norm_eps": 1e-12,
|
| 33 |
+
"max_position_embeddings": 512,
|
| 34 |
+
"model_type": "albert",
|
| 35 |
+
"net_structure_type": 0,
|
| 36 |
+
"num_attention_heads": 12,
|
| 37 |
+
"num_hidden_groups": 1,
|
| 38 |
+
"num_hidden_layers": 12,
|
| 39 |
+
"num_memory_blocks": 0,
|
| 40 |
+
"pad_token_id": 0,
|
| 41 |
+
"position_embedding_type": "absolute",
|
| 42 |
+
"problem_type": "single_label_classification",
|
| 43 |
+
"torch_dtype": "float32",
|
| 44 |
+
"transformers_version": "4.48.0",
|
| 45 |
+
"type_vocab_size": 2,
|
| 46 |
+
"vocab_size": 30000
|
| 47 |
+
}
|
checkpoint-456/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7de090cbcc2fc88c209787820c086224988f028790b65f74b65f19186c33681f
|
| 3 |
+
size 46750064
|
checkpoint-456/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2fb139b88e9ef22a50055fde14377fc7d2466f823165f07f65f4235334a9c763
|
| 3 |
+
size 93515533
|
checkpoint-456/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e62cf6403d0b3a75e60a594e89c6ea29b639c4d39069da9b2cb7f1b2caeb2f7
|
| 3 |
+
size 13990
|
checkpoint-456/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:704cfd08203f995474c39e6e1d8e50b9e297b05822047b677b9d92081ae459e5
|
| 3 |
+
size 1064
|
checkpoint-456/trainer_state.json
ADDED
|
@@ -0,0 +1,3450 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7394976019859314,
|
| 3 |
+
"best_model_checkpoint": "MerlAIn-Base-Albert-005/checkpoint-456",
|
| 4 |
+
"epoch": 12.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 456,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.02631578947368421,
|
| 13 |
+
"grad_norm": 39.9515380859375,
|
| 14 |
+
"learning_rate": 1.0869565217391306e-06,
|
| 15 |
+
"loss": 1.3383,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.05263157894736842,
|
| 20 |
+
"grad_norm": 40.45222091674805,
|
| 21 |
+
"learning_rate": 2.173913043478261e-06,
|
| 22 |
+
"loss": 1.557,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.07894736842105263,
|
| 27 |
+
"grad_norm": 34.01548385620117,
|
| 28 |
+
"learning_rate": 3.2608695652173914e-06,
|
| 29 |
+
"loss": 1.401,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.10526315789473684,
|
| 34 |
+
"grad_norm": 43.126163482666016,
|
| 35 |
+
"learning_rate": 4.347826086956522e-06,
|
| 36 |
+
"loss": 1.4821,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.13157894736842105,
|
| 41 |
+
"grad_norm": 53.70107650756836,
|
| 42 |
+
"learning_rate": 5.4347826086956525e-06,
|
| 43 |
+
"loss": 1.3236,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.15789473684210525,
|
| 48 |
+
"grad_norm": 35.62350845336914,
|
| 49 |
+
"learning_rate": 6.521739130434783e-06,
|
| 50 |
+
"loss": 1.312,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.18421052631578946,
|
| 55 |
+
"grad_norm": 38.48075485229492,
|
| 56 |
+
"learning_rate": 7.608695652173914e-06,
|
| 57 |
+
"loss": 1.3479,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.21052631578947367,
|
| 62 |
+
"grad_norm": 38.26470184326172,
|
| 63 |
+
"learning_rate": 8.695652173913044e-06,
|
| 64 |
+
"loss": 1.3377,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.23684210526315788,
|
| 69 |
+
"grad_norm": 56.54499816894531,
|
| 70 |
+
"learning_rate": 9.782608695652175e-06,
|
| 71 |
+
"loss": 1.3648,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.2631578947368421,
|
| 76 |
+
"grad_norm": 28.90220832824707,
|
| 77 |
+
"learning_rate": 1.0869565217391305e-05,
|
| 78 |
+
"loss": 1.4225,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2894736842105263,
|
| 83 |
+
"grad_norm": 21.76893424987793,
|
| 84 |
+
"learning_rate": 1.1956521739130435e-05,
|
| 85 |
+
"loss": 1.2528,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.3157894736842105,
|
| 90 |
+
"grad_norm": 26.75815773010254,
|
| 91 |
+
"learning_rate": 1.3043478260869566e-05,
|
| 92 |
+
"loss": 1.4086,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.34210526315789475,
|
| 97 |
+
"grad_norm": 41.68316650390625,
|
| 98 |
+
"learning_rate": 1.4130434782608694e-05,
|
| 99 |
+
"loss": 1.29,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.3684210526315789,
|
| 104 |
+
"grad_norm": 21.34151268005371,
|
| 105 |
+
"learning_rate": 1.5217391304347828e-05,
|
| 106 |
+
"loss": 1.3297,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.39473684210526316,
|
| 111 |
+
"grad_norm": 27.759029388427734,
|
| 112 |
+
"learning_rate": 1.630434782608696e-05,
|
| 113 |
+
"loss": 1.3983,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.42105263157894735,
|
| 118 |
+
"grad_norm": 33.36210250854492,
|
| 119 |
+
"learning_rate": 1.739130434782609e-05,
|
| 120 |
+
"loss": 1.344,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.4473684210526316,
|
| 125 |
+
"grad_norm": 24.369775772094727,
|
| 126 |
+
"learning_rate": 1.8478260869565216e-05,
|
| 127 |
+
"loss": 1.3336,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.47368421052631576,
|
| 132 |
+
"grad_norm": 18.205625534057617,
|
| 133 |
+
"learning_rate": 1.956521739130435e-05,
|
| 134 |
+
"loss": 1.273,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.5,
|
| 139 |
+
"grad_norm": 49.468936920166016,
|
| 140 |
+
"learning_rate": 2.065217391304348e-05,
|
| 141 |
+
"loss": 1.2196,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.5263157894736842,
|
| 146 |
+
"grad_norm": 21.179851531982422,
|
| 147 |
+
"learning_rate": 2.173913043478261e-05,
|
| 148 |
+
"loss": 1.2888,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.5526315789473685,
|
| 153 |
+
"grad_norm": 19.813323974609375,
|
| 154 |
+
"learning_rate": 2.282608695652174e-05,
|
| 155 |
+
"loss": 1.2947,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.5789473684210527,
|
| 160 |
+
"grad_norm": 23.896602630615234,
|
| 161 |
+
"learning_rate": 2.391304347826087e-05,
|
| 162 |
+
"loss": 1.363,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.6052631578947368,
|
| 167 |
+
"grad_norm": 31.29140281677246,
|
| 168 |
+
"learning_rate": 2.5e-05,
|
| 169 |
+
"loss": 1.1668,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.631578947368421,
|
| 174 |
+
"grad_norm": 24.209211349487305,
|
| 175 |
+
"learning_rate": 2.608695652173913e-05,
|
| 176 |
+
"loss": 1.3319,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.6578947368421053,
|
| 181 |
+
"grad_norm": 38.2145881652832,
|
| 182 |
+
"learning_rate": 2.7173913043478262e-05,
|
| 183 |
+
"loss": 1.5831,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.6842105263157895,
|
| 188 |
+
"grad_norm": 35.99103927612305,
|
| 189 |
+
"learning_rate": 2.826086956521739e-05,
|
| 190 |
+
"loss": 1.4907,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.7105263157894737,
|
| 195 |
+
"grad_norm": 50.263328552246094,
|
| 196 |
+
"learning_rate": 2.9347826086956526e-05,
|
| 197 |
+
"loss": 1.2867,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.7368421052631579,
|
| 202 |
+
"grad_norm": 29.839618682861328,
|
| 203 |
+
"learning_rate": 3.0434782608695656e-05,
|
| 204 |
+
"loss": 1.1837,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.7631578947368421,
|
| 209 |
+
"grad_norm": 12.5934476852417,
|
| 210 |
+
"learning_rate": 3.152173913043479e-05,
|
| 211 |
+
"loss": 1.2788,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.7894736842105263,
|
| 216 |
+
"grad_norm": 34.49674987792969,
|
| 217 |
+
"learning_rate": 3.260869565217392e-05,
|
| 218 |
+
"loss": 1.1644,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.8157894736842105,
|
| 223 |
+
"grad_norm": 18.00296974182129,
|
| 224 |
+
"learning_rate": 3.369565217391305e-05,
|
| 225 |
+
"loss": 1.2322,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.8421052631578947,
|
| 230 |
+
"grad_norm": 42.88111114501953,
|
| 231 |
+
"learning_rate": 3.478260869565218e-05,
|
| 232 |
+
"loss": 1.392,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.868421052631579,
|
| 237 |
+
"grad_norm": 24.13482093811035,
|
| 238 |
+
"learning_rate": 3.58695652173913e-05,
|
| 239 |
+
"loss": 1.2026,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.8947368421052632,
|
| 244 |
+
"grad_norm": 17.98135757446289,
|
| 245 |
+
"learning_rate": 3.695652173913043e-05,
|
| 246 |
+
"loss": 1.3158,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.9210526315789473,
|
| 251 |
+
"grad_norm": 32.5180778503418,
|
| 252 |
+
"learning_rate": 3.804347826086957e-05,
|
| 253 |
+
"loss": 1.4483,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.9473684210526315,
|
| 258 |
+
"grad_norm": 22.26108169555664,
|
| 259 |
+
"learning_rate": 3.91304347826087e-05,
|
| 260 |
+
"loss": 1.4622,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.9736842105263158,
|
| 265 |
+
"grad_norm": 23.253055572509766,
|
| 266 |
+
"learning_rate": 4.021739130434783e-05,
|
| 267 |
+
"loss": 1.3209,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 1.0,
|
| 272 |
+
"grad_norm": 37.42800521850586,
|
| 273 |
+
"learning_rate": 4.130434782608696e-05,
|
| 274 |
+
"loss": 1.3371,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.0,
|
| 279 |
+
"eval_accuracy": 0.31666666666666665,
|
| 280 |
+
"eval_f1_macro": 0.23189176986584106,
|
| 281 |
+
"eval_f1_micro": 0.31666666666666665,
|
| 282 |
+
"eval_f1_weighted": 0.23189176986584106,
|
| 283 |
+
"eval_loss": 1.3728160858154297,
|
| 284 |
+
"eval_precision_macro": 0.22630718954248366,
|
| 285 |
+
"eval_precision_micro": 0.31666666666666665,
|
| 286 |
+
"eval_precision_weighted": 0.22630718954248366,
|
| 287 |
+
"eval_recall_macro": 0.31666666666666665,
|
| 288 |
+
"eval_recall_micro": 0.31666666666666665,
|
| 289 |
+
"eval_recall_weighted": 0.31666666666666665,
|
| 290 |
+
"eval_runtime": 61.179,
|
| 291 |
+
"eval_samples_per_second": 0.981,
|
| 292 |
+
"eval_steps_per_second": 0.033,
|
| 293 |
+
"step": 38
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.0263157894736843,
|
| 297 |
+
"grad_norm": 37.619869232177734,
|
| 298 |
+
"learning_rate": 4.239130434782609e-05,
|
| 299 |
+
"loss": 1.2901,
|
| 300 |
+
"step": 39
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 1.0526315789473684,
|
| 304 |
+
"grad_norm": 19.8093204498291,
|
| 305 |
+
"learning_rate": 4.347826086956522e-05,
|
| 306 |
+
"loss": 1.2963,
|
| 307 |
+
"step": 40
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 1.0789473684210527,
|
| 311 |
+
"grad_norm": 22.44889259338379,
|
| 312 |
+
"learning_rate": 4.456521739130435e-05,
|
| 313 |
+
"loss": 1.0632,
|
| 314 |
+
"step": 41
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.1052631578947367,
|
| 318 |
+
"grad_norm": 42.281307220458984,
|
| 319 |
+
"learning_rate": 4.565217391304348e-05,
|
| 320 |
+
"loss": 1.3932,
|
| 321 |
+
"step": 42
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 1.131578947368421,
|
| 325 |
+
"grad_norm": 33.52845764160156,
|
| 326 |
+
"learning_rate": 4.673913043478261e-05,
|
| 327 |
+
"loss": 1.2337,
|
| 328 |
+
"step": 43
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 1.1578947368421053,
|
| 332 |
+
"grad_norm": 27.728843688964844,
|
| 333 |
+
"learning_rate": 4.782608695652174e-05,
|
| 334 |
+
"loss": 1.2061,
|
| 335 |
+
"step": 44
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 1.1842105263157894,
|
| 339 |
+
"grad_norm": 41.600250244140625,
|
| 340 |
+
"learning_rate": 4.891304347826087e-05,
|
| 341 |
+
"loss": 1.5733,
|
| 342 |
+
"step": 45
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.2105263157894737,
|
| 346 |
+
"grad_norm": 17.71347427368164,
|
| 347 |
+
"learning_rate": 5e-05,
|
| 348 |
+
"loss": 1.3113,
|
| 349 |
+
"step": 46
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 1.236842105263158,
|
| 353 |
+
"grad_norm": 68.43779754638672,
|
| 354 |
+
"learning_rate": 4.987804878048781e-05,
|
| 355 |
+
"loss": 1.3112,
|
| 356 |
+
"step": 47
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 1.263157894736842,
|
| 360 |
+
"grad_norm": 40.969791412353516,
|
| 361 |
+
"learning_rate": 4.975609756097561e-05,
|
| 362 |
+
"loss": 1.3548,
|
| 363 |
+
"step": 48
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 1.2894736842105263,
|
| 367 |
+
"grad_norm": 30.01605224609375,
|
| 368 |
+
"learning_rate": 4.9634146341463415e-05,
|
| 369 |
+
"loss": 1.2392,
|
| 370 |
+
"step": 49
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 1.3157894736842106,
|
| 374 |
+
"grad_norm": 19.757888793945312,
|
| 375 |
+
"learning_rate": 4.951219512195122e-05,
|
| 376 |
+
"loss": 1.2283,
|
| 377 |
+
"step": 50
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.3421052631578947,
|
| 381 |
+
"grad_norm": 24.616483688354492,
|
| 382 |
+
"learning_rate": 4.9390243902439024e-05,
|
| 383 |
+
"loss": 1.3452,
|
| 384 |
+
"step": 51
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.368421052631579,
|
| 388 |
+
"grad_norm": 41.386173248291016,
|
| 389 |
+
"learning_rate": 4.926829268292683e-05,
|
| 390 |
+
"loss": 1.1968,
|
| 391 |
+
"step": 52
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 1.3947368421052633,
|
| 395 |
+
"grad_norm": 47.946109771728516,
|
| 396 |
+
"learning_rate": 4.914634146341464e-05,
|
| 397 |
+
"loss": 1.0527,
|
| 398 |
+
"step": 53
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.4210526315789473,
|
| 402 |
+
"grad_norm": 15.215785026550293,
|
| 403 |
+
"learning_rate": 4.902439024390244e-05,
|
| 404 |
+
"loss": 1.0679,
|
| 405 |
+
"step": 54
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.4473684210526316,
|
| 409 |
+
"grad_norm": 112.38186645507812,
|
| 410 |
+
"learning_rate": 4.890243902439025e-05,
|
| 411 |
+
"loss": 1.4802,
|
| 412 |
+
"step": 55
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 1.4736842105263157,
|
| 416 |
+
"grad_norm": 44.076480865478516,
|
| 417 |
+
"learning_rate": 4.878048780487805e-05,
|
| 418 |
+
"loss": 1.5318,
|
| 419 |
+
"step": 56
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.5,
|
| 423 |
+
"grad_norm": 28.927885055541992,
|
| 424 |
+
"learning_rate": 4.8658536585365856e-05,
|
| 425 |
+
"loss": 1.3589,
|
| 426 |
+
"step": 57
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 1.526315789473684,
|
| 430 |
+
"grad_norm": 23.650590896606445,
|
| 431 |
+
"learning_rate": 4.853658536585366e-05,
|
| 432 |
+
"loss": 1.2349,
|
| 433 |
+
"step": 58
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 1.5526315789473686,
|
| 437 |
+
"grad_norm": 26.140121459960938,
|
| 438 |
+
"learning_rate": 4.8414634146341464e-05,
|
| 439 |
+
"loss": 1.3551,
|
| 440 |
+
"step": 59
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.5789473684210527,
|
| 444 |
+
"grad_norm": 19.280208587646484,
|
| 445 |
+
"learning_rate": 4.829268292682927e-05,
|
| 446 |
+
"loss": 1.3492,
|
| 447 |
+
"step": 60
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 1.6052631578947367,
|
| 451 |
+
"grad_norm": 13.773977279663086,
|
| 452 |
+
"learning_rate": 4.817073170731707e-05,
|
| 453 |
+
"loss": 1.4615,
|
| 454 |
+
"step": 61
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 1.631578947368421,
|
| 458 |
+
"grad_norm": 23.0042724609375,
|
| 459 |
+
"learning_rate": 4.804878048780488e-05,
|
| 460 |
+
"loss": 1.3364,
|
| 461 |
+
"step": 62
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.6578947368421053,
|
| 465 |
+
"grad_norm": 10.230228424072266,
|
| 466 |
+
"learning_rate": 4.792682926829268e-05,
|
| 467 |
+
"loss": 1.3181,
|
| 468 |
+
"step": 63
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.6842105263157894,
|
| 472 |
+
"grad_norm": 11.699337005615234,
|
| 473 |
+
"learning_rate": 4.7804878048780485e-05,
|
| 474 |
+
"loss": 1.4606,
|
| 475 |
+
"step": 64
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.7105263157894737,
|
| 479 |
+
"grad_norm": 9.87794303894043,
|
| 480 |
+
"learning_rate": 4.7682926829268296e-05,
|
| 481 |
+
"loss": 1.2379,
|
| 482 |
+
"step": 65
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.736842105263158,
|
| 486 |
+
"grad_norm": 15.70111083984375,
|
| 487 |
+
"learning_rate": 4.75609756097561e-05,
|
| 488 |
+
"loss": 1.3018,
|
| 489 |
+
"step": 66
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.763157894736842,
|
| 493 |
+
"grad_norm": 13.613493919372559,
|
| 494 |
+
"learning_rate": 4.7439024390243905e-05,
|
| 495 |
+
"loss": 1.1586,
|
| 496 |
+
"step": 67
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.7894736842105263,
|
| 500 |
+
"grad_norm": 10.26297378540039,
|
| 501 |
+
"learning_rate": 4.731707317073171e-05,
|
| 502 |
+
"loss": 1.3906,
|
| 503 |
+
"step": 68
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.8157894736842106,
|
| 507 |
+
"grad_norm": 19.95726203918457,
|
| 508 |
+
"learning_rate": 4.719512195121951e-05,
|
| 509 |
+
"loss": 1.2608,
|
| 510 |
+
"step": 69
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 1.8421052631578947,
|
| 514 |
+
"grad_norm": 13.270914077758789,
|
| 515 |
+
"learning_rate": 4.707317073170732e-05,
|
| 516 |
+
"loss": 1.4012,
|
| 517 |
+
"step": 70
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 1.868421052631579,
|
| 521 |
+
"grad_norm": 11.952712059020996,
|
| 522 |
+
"learning_rate": 4.695121951219512e-05,
|
| 523 |
+
"loss": 1.051,
|
| 524 |
+
"step": 71
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.8947368421052633,
|
| 528 |
+
"grad_norm": 20.612009048461914,
|
| 529 |
+
"learning_rate": 4.682926829268293e-05,
|
| 530 |
+
"loss": 1.2791,
|
| 531 |
+
"step": 72
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.9210526315789473,
|
| 535 |
+
"grad_norm": 16.857328414916992,
|
| 536 |
+
"learning_rate": 4.670731707317074e-05,
|
| 537 |
+
"loss": 1.3527,
|
| 538 |
+
"step": 73
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.9473684210526314,
|
| 542 |
+
"grad_norm": 10.960662841796875,
|
| 543 |
+
"learning_rate": 4.658536585365854e-05,
|
| 544 |
+
"loss": 1.2018,
|
| 545 |
+
"step": 74
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.973684210526316,
|
| 549 |
+
"grad_norm": 33.91846466064453,
|
| 550 |
+
"learning_rate": 4.6463414634146345e-05,
|
| 551 |
+
"loss": 1.3579,
|
| 552 |
+
"step": 75
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 2.0,
|
| 556 |
+
"grad_norm": 17.503137588500977,
|
| 557 |
+
"learning_rate": 4.634146341463415e-05,
|
| 558 |
+
"loss": 1.219,
|
| 559 |
+
"step": 76
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 2.0,
|
| 563 |
+
"eval_accuracy": 0.25,
|
| 564 |
+
"eval_f1_macro": 0.18048340548340547,
|
| 565 |
+
"eval_f1_micro": 0.25,
|
| 566 |
+
"eval_f1_weighted": 0.1804834054834055,
|
| 567 |
+
"eval_loss": 1.4174046516418457,
|
| 568 |
+
"eval_precision_macro": 0.2144345238095238,
|
| 569 |
+
"eval_precision_micro": 0.25,
|
| 570 |
+
"eval_precision_weighted": 0.2144345238095238,
|
| 571 |
+
"eval_recall_macro": 0.25,
|
| 572 |
+
"eval_recall_micro": 0.25,
|
| 573 |
+
"eval_recall_weighted": 0.25,
|
| 574 |
+
"eval_runtime": 60.4131,
|
| 575 |
+
"eval_samples_per_second": 0.993,
|
| 576 |
+
"eval_steps_per_second": 0.033,
|
| 577 |
+
"step": 76
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 2.026315789473684,
|
| 581 |
+
"grad_norm": 25.20892333984375,
|
| 582 |
+
"learning_rate": 4.6219512195121954e-05,
|
| 583 |
+
"loss": 1.2875,
|
| 584 |
+
"step": 77
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.0526315789473686,
|
| 588 |
+
"grad_norm": 24.98311424255371,
|
| 589 |
+
"learning_rate": 4.609756097560976e-05,
|
| 590 |
+
"loss": 1.1022,
|
| 591 |
+
"step": 78
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.0789473684210527,
|
| 595 |
+
"grad_norm": 25.237098693847656,
|
| 596 |
+
"learning_rate": 4.597560975609756e-05,
|
| 597 |
+
"loss": 1.0668,
|
| 598 |
+
"step": 79
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 2.1052631578947367,
|
| 602 |
+
"grad_norm": 36.29819107055664,
|
| 603 |
+
"learning_rate": 4.585365853658537e-05,
|
| 604 |
+
"loss": 1.3539,
|
| 605 |
+
"step": 80
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 2.1315789473684212,
|
| 609 |
+
"grad_norm": 35.843807220458984,
|
| 610 |
+
"learning_rate": 4.573170731707318e-05,
|
| 611 |
+
"loss": 1.4779,
|
| 612 |
+
"step": 81
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 2.1578947368421053,
|
| 616 |
+
"grad_norm": 112.81155395507812,
|
| 617 |
+
"learning_rate": 4.560975609756098e-05,
|
| 618 |
+
"loss": 1.2617,
|
| 619 |
+
"step": 82
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 2.1842105263157894,
|
| 623 |
+
"grad_norm": 28.52109718322754,
|
| 624 |
+
"learning_rate": 4.5487804878048786e-05,
|
| 625 |
+
"loss": 1.1843,
|
| 626 |
+
"step": 83
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 2.2105263157894735,
|
| 630 |
+
"grad_norm": 24.48183822631836,
|
| 631 |
+
"learning_rate": 4.536585365853659e-05,
|
| 632 |
+
"loss": 1.2935,
|
| 633 |
+
"step": 84
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.236842105263158,
|
| 637 |
+
"grad_norm": 40.971046447753906,
|
| 638 |
+
"learning_rate": 4.5243902439024394e-05,
|
| 639 |
+
"loss": 1.3401,
|
| 640 |
+
"step": 85
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 2.263157894736842,
|
| 644 |
+
"grad_norm": 21.38822364807129,
|
| 645 |
+
"learning_rate": 4.51219512195122e-05,
|
| 646 |
+
"loss": 1.0817,
|
| 647 |
+
"step": 86
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 2.2894736842105265,
|
| 651 |
+
"grad_norm": 20.794170379638672,
|
| 652 |
+
"learning_rate": 4.5e-05,
|
| 653 |
+
"loss": 1.1627,
|
| 654 |
+
"step": 87
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 2.3157894736842106,
|
| 658 |
+
"grad_norm": 21.098356246948242,
|
| 659 |
+
"learning_rate": 4.487804878048781e-05,
|
| 660 |
+
"loss": 1.2119,
|
| 661 |
+
"step": 88
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 2.3421052631578947,
|
| 665 |
+
"grad_norm": 22.802223205566406,
|
| 666 |
+
"learning_rate": 4.475609756097561e-05,
|
| 667 |
+
"loss": 1.2992,
|
| 668 |
+
"step": 89
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 2.3684210526315788,
|
| 672 |
+
"grad_norm": 18.339962005615234,
|
| 673 |
+
"learning_rate": 4.4634146341463416e-05,
|
| 674 |
+
"loss": 1.034,
|
| 675 |
+
"step": 90
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 2.3947368421052633,
|
| 679 |
+
"grad_norm": 53.66396713256836,
|
| 680 |
+
"learning_rate": 4.451219512195122e-05,
|
| 681 |
+
"loss": 1.207,
|
| 682 |
+
"step": 91
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 2.4210526315789473,
|
| 686 |
+
"grad_norm": 19.618972778320312,
|
| 687 |
+
"learning_rate": 4.4390243902439024e-05,
|
| 688 |
+
"loss": 1.1393,
|
| 689 |
+
"step": 92
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 2.4473684210526314,
|
| 693 |
+
"grad_norm": 21.265857696533203,
|
| 694 |
+
"learning_rate": 4.4268292682926835e-05,
|
| 695 |
+
"loss": 1.1223,
|
| 696 |
+
"step": 93
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 2.473684210526316,
|
| 700 |
+
"grad_norm": 23.079221725463867,
|
| 701 |
+
"learning_rate": 4.414634146341464e-05,
|
| 702 |
+
"loss": 1.0722,
|
| 703 |
+
"step": 94
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 2.5,
|
| 707 |
+
"grad_norm": 21.901897430419922,
|
| 708 |
+
"learning_rate": 4.4024390243902443e-05,
|
| 709 |
+
"loss": 1.0003,
|
| 710 |
+
"step": 95
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 2.526315789473684,
|
| 714 |
+
"grad_norm": 25.181835174560547,
|
| 715 |
+
"learning_rate": 4.390243902439025e-05,
|
| 716 |
+
"loss": 1.0741,
|
| 717 |
+
"step": 96
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 2.5526315789473686,
|
| 721 |
+
"grad_norm": 31.18767547607422,
|
| 722 |
+
"learning_rate": 4.378048780487805e-05,
|
| 723 |
+
"loss": 1.0066,
|
| 724 |
+
"step": 97
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 2.5789473684210527,
|
| 728 |
+
"grad_norm": 60.933631896972656,
|
| 729 |
+
"learning_rate": 4.3658536585365856e-05,
|
| 730 |
+
"loss": 1.2893,
|
| 731 |
+
"step": 98
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 2.6052631578947367,
|
| 735 |
+
"grad_norm": 23.306636810302734,
|
| 736 |
+
"learning_rate": 4.353658536585366e-05,
|
| 737 |
+
"loss": 1.2187,
|
| 738 |
+
"step": 99
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 2.6315789473684212,
|
| 742 |
+
"grad_norm": 30.128135681152344,
|
| 743 |
+
"learning_rate": 4.3414634146341465e-05,
|
| 744 |
+
"loss": 0.9419,
|
| 745 |
+
"step": 100
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 2.6578947368421053,
|
| 749 |
+
"grad_norm": 28.193851470947266,
|
| 750 |
+
"learning_rate": 4.329268292682927e-05,
|
| 751 |
+
"loss": 1.0172,
|
| 752 |
+
"step": 101
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 2.6842105263157894,
|
| 756 |
+
"grad_norm": 54.759422302246094,
|
| 757 |
+
"learning_rate": 4.317073170731707e-05,
|
| 758 |
+
"loss": 1.2784,
|
| 759 |
+
"step": 102
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 2.7105263157894735,
|
| 763 |
+
"grad_norm": 32.32737731933594,
|
| 764 |
+
"learning_rate": 4.304878048780488e-05,
|
| 765 |
+
"loss": 1.7073,
|
| 766 |
+
"step": 103
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 2.736842105263158,
|
| 770 |
+
"grad_norm": 20.45136260986328,
|
| 771 |
+
"learning_rate": 4.292682926829268e-05,
|
| 772 |
+
"loss": 0.8336,
|
| 773 |
+
"step": 104
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 2.763157894736842,
|
| 777 |
+
"grad_norm": 25.78130531311035,
|
| 778 |
+
"learning_rate": 4.2804878048780486e-05,
|
| 779 |
+
"loss": 0.9388,
|
| 780 |
+
"step": 105
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 2.7894736842105265,
|
| 784 |
+
"grad_norm": 44.98958206176758,
|
| 785 |
+
"learning_rate": 4.26829268292683e-05,
|
| 786 |
+
"loss": 1.0452,
|
| 787 |
+
"step": 106
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 2.8157894736842106,
|
| 791 |
+
"grad_norm": 25.93155288696289,
|
| 792 |
+
"learning_rate": 4.25609756097561e-05,
|
| 793 |
+
"loss": 1.7353,
|
| 794 |
+
"step": 107
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 2.8421052631578947,
|
| 798 |
+
"grad_norm": 39.97303771972656,
|
| 799 |
+
"learning_rate": 4.2439024390243905e-05,
|
| 800 |
+
"loss": 1.0937,
|
| 801 |
+
"step": 108
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.8684210526315788,
|
| 805 |
+
"grad_norm": 28.07468032836914,
|
| 806 |
+
"learning_rate": 4.231707317073171e-05,
|
| 807 |
+
"loss": 1.0547,
|
| 808 |
+
"step": 109
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 2.8947368421052633,
|
| 812 |
+
"grad_norm": 33.94812774658203,
|
| 813 |
+
"learning_rate": 4.2195121951219514e-05,
|
| 814 |
+
"loss": 1.4567,
|
| 815 |
+
"step": 110
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 2.9210526315789473,
|
| 819 |
+
"grad_norm": 19.667274475097656,
|
| 820 |
+
"learning_rate": 4.207317073170732e-05,
|
| 821 |
+
"loss": 0.9645,
|
| 822 |
+
"step": 111
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 2.9473684210526314,
|
| 826 |
+
"grad_norm": 22.64081382751465,
|
| 827 |
+
"learning_rate": 4.195121951219512e-05,
|
| 828 |
+
"loss": 1.0789,
|
| 829 |
+
"step": 112
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 2.973684210526316,
|
| 833 |
+
"grad_norm": 31.188119888305664,
|
| 834 |
+
"learning_rate": 4.1829268292682926e-05,
|
| 835 |
+
"loss": 1.1605,
|
| 836 |
+
"step": 113
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 3.0,
|
| 840 |
+
"grad_norm": 22.26033592224121,
|
| 841 |
+
"learning_rate": 4.170731707317073e-05,
|
| 842 |
+
"loss": 1.0198,
|
| 843 |
+
"step": 114
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 3.0,
|
| 847 |
+
"eval_accuracy": 0.31666666666666665,
|
| 848 |
+
"eval_f1_macro": 0.30134680134680136,
|
| 849 |
+
"eval_f1_micro": 0.31666666666666665,
|
| 850 |
+
"eval_f1_weighted": 0.3013468013468013,
|
| 851 |
+
"eval_loss": 1.318847417831421,
|
| 852 |
+
"eval_precision_macro": 0.306547619047619,
|
| 853 |
+
"eval_precision_micro": 0.31666666666666665,
|
| 854 |
+
"eval_precision_weighted": 0.30654761904761896,
|
| 855 |
+
"eval_recall_macro": 0.31666666666666665,
|
| 856 |
+
"eval_recall_micro": 0.31666666666666665,
|
| 857 |
+
"eval_recall_weighted": 0.31666666666666665,
|
| 858 |
+
"eval_runtime": 66.5323,
|
| 859 |
+
"eval_samples_per_second": 0.902,
|
| 860 |
+
"eval_steps_per_second": 0.03,
|
| 861 |
+
"step": 114
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 3.026315789473684,
|
| 865 |
+
"grad_norm": 22.01658821105957,
|
| 866 |
+
"learning_rate": 4.1585365853658535e-05,
|
| 867 |
+
"loss": 0.9237,
|
| 868 |
+
"step": 115
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 3.0526315789473686,
|
| 872 |
+
"grad_norm": 34.970157623291016,
|
| 873 |
+
"learning_rate": 4.146341463414634e-05,
|
| 874 |
+
"loss": 1.1732,
|
| 875 |
+
"step": 116
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 3.0789473684210527,
|
| 879 |
+
"grad_norm": 38.19720458984375,
|
| 880 |
+
"learning_rate": 4.134146341463414e-05,
|
| 881 |
+
"loss": 0.9921,
|
| 882 |
+
"step": 117
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 3.1052631578947367,
|
| 886 |
+
"grad_norm": 12.365692138671875,
|
| 887 |
+
"learning_rate": 4.1219512195121954e-05,
|
| 888 |
+
"loss": 0.7438,
|
| 889 |
+
"step": 118
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 3.1315789473684212,
|
| 893 |
+
"grad_norm": 11.447243690490723,
|
| 894 |
+
"learning_rate": 4.109756097560976e-05,
|
| 895 |
+
"loss": 0.8613,
|
| 896 |
+
"step": 119
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 3.1578947368421053,
|
| 900 |
+
"grad_norm": 23.869182586669922,
|
| 901 |
+
"learning_rate": 4.097560975609756e-05,
|
| 902 |
+
"loss": 0.9255,
|
| 903 |
+
"step": 120
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 3.1842105263157894,
|
| 907 |
+
"grad_norm": 23.761123657226562,
|
| 908 |
+
"learning_rate": 4.085365853658537e-05,
|
| 909 |
+
"loss": 1.0873,
|
| 910 |
+
"step": 121
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 3.2105263157894735,
|
| 914 |
+
"grad_norm": 49.23319625854492,
|
| 915 |
+
"learning_rate": 4.073170731707317e-05,
|
| 916 |
+
"loss": 1.2417,
|
| 917 |
+
"step": 122
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 3.236842105263158,
|
| 921 |
+
"grad_norm": 20.207958221435547,
|
| 922 |
+
"learning_rate": 4.060975609756098e-05,
|
| 923 |
+
"loss": 0.9844,
|
| 924 |
+
"step": 123
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 3.263157894736842,
|
| 928 |
+
"grad_norm": 27.667728424072266,
|
| 929 |
+
"learning_rate": 4.0487804878048786e-05,
|
| 930 |
+
"loss": 0.9163,
|
| 931 |
+
"step": 124
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 3.2894736842105265,
|
| 935 |
+
"grad_norm": 26.463153839111328,
|
| 936 |
+
"learning_rate": 4.036585365853659e-05,
|
| 937 |
+
"loss": 0.992,
|
| 938 |
+
"step": 125
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 3.3157894736842106,
|
| 942 |
+
"grad_norm": 30.33213996887207,
|
| 943 |
+
"learning_rate": 4.0243902439024395e-05,
|
| 944 |
+
"loss": 0.9267,
|
| 945 |
+
"step": 126
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 3.3421052631578947,
|
| 949 |
+
"grad_norm": 25.357423782348633,
|
| 950 |
+
"learning_rate": 4.01219512195122e-05,
|
| 951 |
+
"loss": 1.0649,
|
| 952 |
+
"step": 127
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 3.3684210526315788,
|
| 956 |
+
"grad_norm": 32.58332443237305,
|
| 957 |
+
"learning_rate": 4e-05,
|
| 958 |
+
"loss": 1.3273,
|
| 959 |
+
"step": 128
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 3.3947368421052633,
|
| 963 |
+
"grad_norm": 27.008098602294922,
|
| 964 |
+
"learning_rate": 3.987804878048781e-05,
|
| 965 |
+
"loss": 0.7744,
|
| 966 |
+
"step": 129
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 3.4210526315789473,
|
| 970 |
+
"grad_norm": 28.078779220581055,
|
| 971 |
+
"learning_rate": 3.975609756097561e-05,
|
| 972 |
+
"loss": 0.7122,
|
| 973 |
+
"step": 130
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 3.4473684210526314,
|
| 977 |
+
"grad_norm": 42.70326232910156,
|
| 978 |
+
"learning_rate": 3.9634146341463416e-05,
|
| 979 |
+
"loss": 1.0784,
|
| 980 |
+
"step": 131
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 3.473684210526316,
|
| 984 |
+
"grad_norm": 49.816429138183594,
|
| 985 |
+
"learning_rate": 3.951219512195122e-05,
|
| 986 |
+
"loss": 0.8521,
|
| 987 |
+
"step": 132
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 3.5,
|
| 991 |
+
"grad_norm": 49.62358474731445,
|
| 992 |
+
"learning_rate": 3.939024390243903e-05,
|
| 993 |
+
"loss": 1.0135,
|
| 994 |
+
"step": 133
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 3.526315789473684,
|
| 998 |
+
"grad_norm": 29.133995056152344,
|
| 999 |
+
"learning_rate": 3.9268292682926835e-05,
|
| 1000 |
+
"loss": 0.9807,
|
| 1001 |
+
"step": 134
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 3.5526315789473686,
|
| 1005 |
+
"grad_norm": 38.17136001586914,
|
| 1006 |
+
"learning_rate": 3.914634146341464e-05,
|
| 1007 |
+
"loss": 1.0422,
|
| 1008 |
+
"step": 135
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 3.5789473684210527,
|
| 1012 |
+
"grad_norm": 37.97235870361328,
|
| 1013 |
+
"learning_rate": 3.9024390243902444e-05,
|
| 1014 |
+
"loss": 0.7642,
|
| 1015 |
+
"step": 136
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 3.6052631578947367,
|
| 1019 |
+
"grad_norm": 48.906150817871094,
|
| 1020 |
+
"learning_rate": 3.890243902439025e-05,
|
| 1021 |
+
"loss": 1.164,
|
| 1022 |
+
"step": 137
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 3.6315789473684212,
|
| 1026 |
+
"grad_norm": 29.808584213256836,
|
| 1027 |
+
"learning_rate": 3.878048780487805e-05,
|
| 1028 |
+
"loss": 0.9559,
|
| 1029 |
+
"step": 138
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 3.6578947368421053,
|
| 1033 |
+
"grad_norm": 34.88420867919922,
|
| 1034 |
+
"learning_rate": 3.8658536585365857e-05,
|
| 1035 |
+
"loss": 1.3472,
|
| 1036 |
+
"step": 139
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 3.6842105263157894,
|
| 1040 |
+
"grad_norm": 20.547361373901367,
|
| 1041 |
+
"learning_rate": 3.853658536585366e-05,
|
| 1042 |
+
"loss": 0.7452,
|
| 1043 |
+
"step": 140
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 3.7105263157894735,
|
| 1047 |
+
"grad_norm": 53.17014694213867,
|
| 1048 |
+
"learning_rate": 3.8414634146341465e-05,
|
| 1049 |
+
"loss": 1.1893,
|
| 1050 |
+
"step": 141
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 3.736842105263158,
|
| 1054 |
+
"grad_norm": 36.8974723815918,
|
| 1055 |
+
"learning_rate": 3.829268292682927e-05,
|
| 1056 |
+
"loss": 1.1465,
|
| 1057 |
+
"step": 142
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 3.763157894736842,
|
| 1061 |
+
"grad_norm": 79.12947082519531,
|
| 1062 |
+
"learning_rate": 3.8170731707317073e-05,
|
| 1063 |
+
"loss": 1.1274,
|
| 1064 |
+
"step": 143
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 3.7894736842105265,
|
| 1068 |
+
"grad_norm": 46.056846618652344,
|
| 1069 |
+
"learning_rate": 3.804878048780488e-05,
|
| 1070 |
+
"loss": 0.8157,
|
| 1071 |
+
"step": 144
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 3.8157894736842106,
|
| 1075 |
+
"grad_norm": 52.49138641357422,
|
| 1076 |
+
"learning_rate": 3.792682926829268e-05,
|
| 1077 |
+
"loss": 0.923,
|
| 1078 |
+
"step": 145
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 3.8421052631578947,
|
| 1082 |
+
"grad_norm": 142.27134704589844,
|
| 1083 |
+
"learning_rate": 3.780487804878049e-05,
|
| 1084 |
+
"loss": 1.2777,
|
| 1085 |
+
"step": 146
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 3.8684210526315788,
|
| 1089 |
+
"grad_norm": 172.19480895996094,
|
| 1090 |
+
"learning_rate": 3.76829268292683e-05,
|
| 1091 |
+
"loss": 1.0432,
|
| 1092 |
+
"step": 147
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 3.8947368421052633,
|
| 1096 |
+
"grad_norm": 123.80152130126953,
|
| 1097 |
+
"learning_rate": 3.75609756097561e-05,
|
| 1098 |
+
"loss": 0.7937,
|
| 1099 |
+
"step": 148
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 3.9210526315789473,
|
| 1103 |
+
"grad_norm": 77.49285125732422,
|
| 1104 |
+
"learning_rate": 3.7439024390243906e-05,
|
| 1105 |
+
"loss": 0.6857,
|
| 1106 |
+
"step": 149
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 3.9473684210526314,
|
| 1110 |
+
"grad_norm": 82.35449981689453,
|
| 1111 |
+
"learning_rate": 3.731707317073171e-05,
|
| 1112 |
+
"loss": 0.9595,
|
| 1113 |
+
"step": 150
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 3.973684210526316,
|
| 1117 |
+
"grad_norm": 38.3079833984375,
|
| 1118 |
+
"learning_rate": 3.7195121951219514e-05,
|
| 1119 |
+
"loss": 0.7892,
|
| 1120 |
+
"step": 151
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 4.0,
|
| 1124 |
+
"grad_norm": 119.38838195800781,
|
| 1125 |
+
"learning_rate": 3.707317073170732e-05,
|
| 1126 |
+
"loss": 1.4254,
|
| 1127 |
+
"step": 152
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 4.0,
|
| 1131 |
+
"eval_accuracy": 0.45,
|
| 1132 |
+
"eval_f1_macro": 0.4606509735541993,
|
| 1133 |
+
"eval_f1_micro": 0.45,
|
| 1134 |
+
"eval_f1_weighted": 0.4606509735541993,
|
| 1135 |
+
"eval_loss": 1.090307354927063,
|
| 1136 |
+
"eval_precision_macro": 0.4900568181818182,
|
| 1137 |
+
"eval_precision_micro": 0.45,
|
| 1138 |
+
"eval_precision_weighted": 0.4900568181818182,
|
| 1139 |
+
"eval_recall_macro": 0.44999999999999996,
|
| 1140 |
+
"eval_recall_micro": 0.45,
|
| 1141 |
+
"eval_recall_weighted": 0.45,
|
| 1142 |
+
"eval_runtime": 69.1646,
|
| 1143 |
+
"eval_samples_per_second": 0.867,
|
| 1144 |
+
"eval_steps_per_second": 0.029,
|
| 1145 |
+
"step": 152
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 4.026315789473684,
|
| 1149 |
+
"grad_norm": 91.3956527709961,
|
| 1150 |
+
"learning_rate": 3.695121951219512e-05,
|
| 1151 |
+
"loss": 1.0411,
|
| 1152 |
+
"step": 153
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 4.052631578947368,
|
| 1156 |
+
"grad_norm": 57.57364273071289,
|
| 1157 |
+
"learning_rate": 3.682926829268293e-05,
|
| 1158 |
+
"loss": 0.7083,
|
| 1159 |
+
"step": 154
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 4.078947368421052,
|
| 1163 |
+
"grad_norm": 82.79764556884766,
|
| 1164 |
+
"learning_rate": 3.670731707317073e-05,
|
| 1165 |
+
"loss": 0.9112,
|
| 1166 |
+
"step": 155
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 4.105263157894737,
|
| 1170 |
+
"grad_norm": 47.4633674621582,
|
| 1171 |
+
"learning_rate": 3.6585365853658535e-05,
|
| 1172 |
+
"loss": 0.9273,
|
| 1173 |
+
"step": 156
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 4.131578947368421,
|
| 1177 |
+
"grad_norm": 59.94166564941406,
|
| 1178 |
+
"learning_rate": 3.646341463414634e-05,
|
| 1179 |
+
"loss": 0.8346,
|
| 1180 |
+
"step": 157
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 4.157894736842105,
|
| 1184 |
+
"grad_norm": 59.959259033203125,
|
| 1185 |
+
"learning_rate": 3.634146341463415e-05,
|
| 1186 |
+
"loss": 0.7768,
|
| 1187 |
+
"step": 158
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 4.184210526315789,
|
| 1191 |
+
"grad_norm": 38.368404388427734,
|
| 1192 |
+
"learning_rate": 3.6219512195121955e-05,
|
| 1193 |
+
"loss": 0.6442,
|
| 1194 |
+
"step": 159
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 4.2105263157894735,
|
| 1198 |
+
"grad_norm": 152.19378662109375,
|
| 1199 |
+
"learning_rate": 3.609756097560976e-05,
|
| 1200 |
+
"loss": 1.1892,
|
| 1201 |
+
"step": 160
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 4.2368421052631575,
|
| 1205 |
+
"grad_norm": 70.09424591064453,
|
| 1206 |
+
"learning_rate": 3.597560975609756e-05,
|
| 1207 |
+
"loss": 0.9863,
|
| 1208 |
+
"step": 161
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 4.2631578947368425,
|
| 1212 |
+
"grad_norm": 35.991065979003906,
|
| 1213 |
+
"learning_rate": 3.585365853658537e-05,
|
| 1214 |
+
"loss": 0.8468,
|
| 1215 |
+
"step": 162
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 4.2894736842105265,
|
| 1219 |
+
"grad_norm": 39.37136459350586,
|
| 1220 |
+
"learning_rate": 3.573170731707317e-05,
|
| 1221 |
+
"loss": 0.8338,
|
| 1222 |
+
"step": 163
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 4.315789473684211,
|
| 1226 |
+
"grad_norm": 103.584716796875,
|
| 1227 |
+
"learning_rate": 3.5609756097560976e-05,
|
| 1228 |
+
"loss": 1.0,
|
| 1229 |
+
"step": 164
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 4.342105263157895,
|
| 1233 |
+
"grad_norm": 38.11565399169922,
|
| 1234 |
+
"learning_rate": 3.548780487804878e-05,
|
| 1235 |
+
"loss": 0.671,
|
| 1236 |
+
"step": 165
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 4.368421052631579,
|
| 1240 |
+
"grad_norm": 45.16068649291992,
|
| 1241 |
+
"learning_rate": 3.5365853658536584e-05,
|
| 1242 |
+
"loss": 0.9203,
|
| 1243 |
+
"step": 166
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 4.394736842105263,
|
| 1247 |
+
"grad_norm": 123.4576416015625,
|
| 1248 |
+
"learning_rate": 3.524390243902439e-05,
|
| 1249 |
+
"loss": 0.9496,
|
| 1250 |
+
"step": 167
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 4.421052631578947,
|
| 1254 |
+
"grad_norm": 33.224815368652344,
|
| 1255 |
+
"learning_rate": 3.512195121951219e-05,
|
| 1256 |
+
"loss": 0.6605,
|
| 1257 |
+
"step": 168
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 4.447368421052632,
|
| 1261 |
+
"grad_norm": 42.245147705078125,
|
| 1262 |
+
"learning_rate": 3.5e-05,
|
| 1263 |
+
"loss": 0.733,
|
| 1264 |
+
"step": 169
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 4.473684210526316,
|
| 1268 |
+
"grad_norm": 35.62926483154297,
|
| 1269 |
+
"learning_rate": 3.48780487804878e-05,
|
| 1270 |
+
"loss": 0.9771,
|
| 1271 |
+
"step": 170
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 4.5,
|
| 1275 |
+
"grad_norm": 36.670928955078125,
|
| 1276 |
+
"learning_rate": 3.475609756097561e-05,
|
| 1277 |
+
"loss": 0.5183,
|
| 1278 |
+
"step": 171
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 4.526315789473684,
|
| 1282 |
+
"grad_norm": 25.697141647338867,
|
| 1283 |
+
"learning_rate": 3.4634146341463416e-05,
|
| 1284 |
+
"loss": 0.8279,
|
| 1285 |
+
"step": 172
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 4.552631578947368,
|
| 1289 |
+
"grad_norm": 50.771846771240234,
|
| 1290 |
+
"learning_rate": 3.451219512195122e-05,
|
| 1291 |
+
"loss": 0.6015,
|
| 1292 |
+
"step": 173
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 4.578947368421053,
|
| 1296 |
+
"grad_norm": 60.814430236816406,
|
| 1297 |
+
"learning_rate": 3.4390243902439025e-05,
|
| 1298 |
+
"loss": 0.7535,
|
| 1299 |
+
"step": 174
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 4.605263157894737,
|
| 1303 |
+
"grad_norm": 29.89155387878418,
|
| 1304 |
+
"learning_rate": 3.4268292682926836e-05,
|
| 1305 |
+
"loss": 0.628,
|
| 1306 |
+
"step": 175
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 4.631578947368421,
|
| 1310 |
+
"grad_norm": 20.725378036499023,
|
| 1311 |
+
"learning_rate": 3.414634146341464e-05,
|
| 1312 |
+
"loss": 0.4225,
|
| 1313 |
+
"step": 176
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 4.657894736842105,
|
| 1317 |
+
"grad_norm": 26.378557205200195,
|
| 1318 |
+
"learning_rate": 3.4024390243902444e-05,
|
| 1319 |
+
"loss": 0.8201,
|
| 1320 |
+
"step": 177
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 4.684210526315789,
|
| 1324 |
+
"grad_norm": 51.66449737548828,
|
| 1325 |
+
"learning_rate": 3.390243902439025e-05,
|
| 1326 |
+
"loss": 0.7664,
|
| 1327 |
+
"step": 178
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 4.7105263157894735,
|
| 1331 |
+
"grad_norm": 47.04916000366211,
|
| 1332 |
+
"learning_rate": 3.378048780487805e-05,
|
| 1333 |
+
"loss": 0.9299,
|
| 1334 |
+
"step": 179
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 4.7368421052631575,
|
| 1338 |
+
"grad_norm": 81.7121353149414,
|
| 1339 |
+
"learning_rate": 3.365853658536586e-05,
|
| 1340 |
+
"loss": 0.8565,
|
| 1341 |
+
"step": 180
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 4.7631578947368425,
|
| 1345 |
+
"grad_norm": 56.56669998168945,
|
| 1346 |
+
"learning_rate": 3.353658536585366e-05,
|
| 1347 |
+
"loss": 0.7974,
|
| 1348 |
+
"step": 181
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 4.7894736842105265,
|
| 1352 |
+
"grad_norm": 44.86998748779297,
|
| 1353 |
+
"learning_rate": 3.3414634146341465e-05,
|
| 1354 |
+
"loss": 0.5811,
|
| 1355 |
+
"step": 182
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 4.815789473684211,
|
| 1359 |
+
"grad_norm": 27.705507278442383,
|
| 1360 |
+
"learning_rate": 3.329268292682927e-05,
|
| 1361 |
+
"loss": 0.6238,
|
| 1362 |
+
"step": 183
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 4.842105263157895,
|
| 1366 |
+
"grad_norm": 27.237136840820312,
|
| 1367 |
+
"learning_rate": 3.3170731707317074e-05,
|
| 1368 |
+
"loss": 0.6288,
|
| 1369 |
+
"step": 184
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 4.868421052631579,
|
| 1373 |
+
"grad_norm": 56.495155334472656,
|
| 1374 |
+
"learning_rate": 3.304878048780488e-05,
|
| 1375 |
+
"loss": 0.6755,
|
| 1376 |
+
"step": 185
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 4.894736842105263,
|
| 1380 |
+
"grad_norm": 51.76018524169922,
|
| 1381 |
+
"learning_rate": 3.292682926829269e-05,
|
| 1382 |
+
"loss": 0.6209,
|
| 1383 |
+
"step": 186
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 4.921052631578947,
|
| 1387 |
+
"grad_norm": 42.66984558105469,
|
| 1388 |
+
"learning_rate": 3.280487804878049e-05,
|
| 1389 |
+
"loss": 0.8452,
|
| 1390 |
+
"step": 187
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 4.947368421052632,
|
| 1394 |
+
"grad_norm": 98.05541229248047,
|
| 1395 |
+
"learning_rate": 3.26829268292683e-05,
|
| 1396 |
+
"loss": 0.8141,
|
| 1397 |
+
"step": 188
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 4.973684210526316,
|
| 1401 |
+
"grad_norm": 43.486305236816406,
|
| 1402 |
+
"learning_rate": 3.25609756097561e-05,
|
| 1403 |
+
"loss": 0.8128,
|
| 1404 |
+
"step": 189
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 5.0,
|
| 1408 |
+
"grad_norm": 48.97419357299805,
|
| 1409 |
+
"learning_rate": 3.2439024390243906e-05,
|
| 1410 |
+
"loss": 0.6552,
|
| 1411 |
+
"step": 190
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 5.0,
|
| 1415 |
+
"eval_accuracy": 0.55,
|
| 1416 |
+
"eval_f1_macro": 0.539948914431673,
|
| 1417 |
+
"eval_f1_micro": 0.55,
|
| 1418 |
+
"eval_f1_weighted": 0.539948914431673,
|
| 1419 |
+
"eval_loss": 0.9649682641029358,
|
| 1420 |
+
"eval_precision_macro": 0.5645833333333333,
|
| 1421 |
+
"eval_precision_micro": 0.55,
|
| 1422 |
+
"eval_precision_weighted": 0.5645833333333333,
|
| 1423 |
+
"eval_recall_macro": 0.55,
|
| 1424 |
+
"eval_recall_micro": 0.55,
|
| 1425 |
+
"eval_recall_weighted": 0.55,
|
| 1426 |
+
"eval_runtime": 71.5473,
|
| 1427 |
+
"eval_samples_per_second": 0.839,
|
| 1428 |
+
"eval_steps_per_second": 0.028,
|
| 1429 |
+
"step": 190
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 5.026315789473684,
|
| 1433 |
+
"grad_norm": 47.699520111083984,
|
| 1434 |
+
"learning_rate": 3.231707317073171e-05,
|
| 1435 |
+
"loss": 0.5292,
|
| 1436 |
+
"step": 191
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 5.052631578947368,
|
| 1440 |
+
"grad_norm": 27.44911003112793,
|
| 1441 |
+
"learning_rate": 3.2195121951219514e-05,
|
| 1442 |
+
"loss": 0.405,
|
| 1443 |
+
"step": 192
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 5.078947368421052,
|
| 1447 |
+
"grad_norm": 35.62791442871094,
|
| 1448 |
+
"learning_rate": 3.207317073170732e-05,
|
| 1449 |
+
"loss": 0.3673,
|
| 1450 |
+
"step": 193
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 5.105263157894737,
|
| 1454 |
+
"grad_norm": 28.489892959594727,
|
| 1455 |
+
"learning_rate": 3.195121951219512e-05,
|
| 1456 |
+
"loss": 0.6755,
|
| 1457 |
+
"step": 194
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 5.131578947368421,
|
| 1461 |
+
"grad_norm": 136.23341369628906,
|
| 1462 |
+
"learning_rate": 3.182926829268293e-05,
|
| 1463 |
+
"loss": 0.774,
|
| 1464 |
+
"step": 195
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 5.157894736842105,
|
| 1468 |
+
"grad_norm": 32.27811813354492,
|
| 1469 |
+
"learning_rate": 3.170731707317073e-05,
|
| 1470 |
+
"loss": 0.4562,
|
| 1471 |
+
"step": 196
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 5.184210526315789,
|
| 1475 |
+
"grad_norm": 43.724117279052734,
|
| 1476 |
+
"learning_rate": 3.1585365853658536e-05,
|
| 1477 |
+
"loss": 0.582,
|
| 1478 |
+
"step": 197
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 5.2105263157894735,
|
| 1482 |
+
"grad_norm": 122.9839096069336,
|
| 1483 |
+
"learning_rate": 3.146341463414634e-05,
|
| 1484 |
+
"loss": 0.8316,
|
| 1485 |
+
"step": 198
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 5.2368421052631575,
|
| 1489 |
+
"grad_norm": 43.039146423339844,
|
| 1490 |
+
"learning_rate": 3.134146341463415e-05,
|
| 1491 |
+
"loss": 0.7792,
|
| 1492 |
+
"step": 199
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 5.2631578947368425,
|
| 1496 |
+
"grad_norm": 48.31818389892578,
|
| 1497 |
+
"learning_rate": 3.1219512195121955e-05,
|
| 1498 |
+
"loss": 0.3813,
|
| 1499 |
+
"step": 200
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 5.2894736842105265,
|
| 1503 |
+
"grad_norm": 27.5224666595459,
|
| 1504 |
+
"learning_rate": 3.109756097560976e-05,
|
| 1505 |
+
"loss": 0.4812,
|
| 1506 |
+
"step": 201
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 5.315789473684211,
|
| 1510 |
+
"grad_norm": 36.04642868041992,
|
| 1511 |
+
"learning_rate": 3.0975609756097564e-05,
|
| 1512 |
+
"loss": 0.6757,
|
| 1513 |
+
"step": 202
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 5.342105263157895,
|
| 1517 |
+
"grad_norm": 33.89840316772461,
|
| 1518 |
+
"learning_rate": 3.085365853658537e-05,
|
| 1519 |
+
"loss": 0.3825,
|
| 1520 |
+
"step": 203
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 5.368421052631579,
|
| 1524 |
+
"grad_norm": 85.91902923583984,
|
| 1525 |
+
"learning_rate": 3.073170731707317e-05,
|
| 1526 |
+
"loss": 0.7955,
|
| 1527 |
+
"step": 204
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 5.394736842105263,
|
| 1531 |
+
"grad_norm": 66.69794464111328,
|
| 1532 |
+
"learning_rate": 3.0609756097560976e-05,
|
| 1533 |
+
"loss": 0.5711,
|
| 1534 |
+
"step": 205
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 5.421052631578947,
|
| 1538 |
+
"grad_norm": 55.40018844604492,
|
| 1539 |
+
"learning_rate": 3.048780487804878e-05,
|
| 1540 |
+
"loss": 0.5853,
|
| 1541 |
+
"step": 206
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 5.447368421052632,
|
| 1545 |
+
"grad_norm": 30.135583877563477,
|
| 1546 |
+
"learning_rate": 3.0365853658536585e-05,
|
| 1547 |
+
"loss": 0.582,
|
| 1548 |
+
"step": 207
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 5.473684210526316,
|
| 1552 |
+
"grad_norm": 31.25365447998047,
|
| 1553 |
+
"learning_rate": 3.0243902439024392e-05,
|
| 1554 |
+
"loss": 0.9509,
|
| 1555 |
+
"step": 208
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 5.5,
|
| 1559 |
+
"grad_norm": 22.51866340637207,
|
| 1560 |
+
"learning_rate": 3.0121951219512197e-05,
|
| 1561 |
+
"loss": 0.273,
|
| 1562 |
+
"step": 209
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 5.526315789473684,
|
| 1566 |
+
"grad_norm": 103.33230590820312,
|
| 1567 |
+
"learning_rate": 3e-05,
|
| 1568 |
+
"loss": 0.5085,
|
| 1569 |
+
"step": 210
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 5.552631578947368,
|
| 1573 |
+
"grad_norm": 49.145145416259766,
|
| 1574 |
+
"learning_rate": 2.9878048780487805e-05,
|
| 1575 |
+
"loss": 0.3109,
|
| 1576 |
+
"step": 211
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 5.578947368421053,
|
| 1580 |
+
"grad_norm": 42.34318923950195,
|
| 1581 |
+
"learning_rate": 2.975609756097561e-05,
|
| 1582 |
+
"loss": 0.5871,
|
| 1583 |
+
"step": 212
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 5.605263157894737,
|
| 1587 |
+
"grad_norm": 42.719451904296875,
|
| 1588 |
+
"learning_rate": 2.9634146341463413e-05,
|
| 1589 |
+
"loss": 0.5125,
|
| 1590 |
+
"step": 213
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 5.631578947368421,
|
| 1594 |
+
"grad_norm": 52.85165786743164,
|
| 1595 |
+
"learning_rate": 2.951219512195122e-05,
|
| 1596 |
+
"loss": 0.8334,
|
| 1597 |
+
"step": 214
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 5.657894736842105,
|
| 1601 |
+
"grad_norm": 103.75800323486328,
|
| 1602 |
+
"learning_rate": 2.9390243902439025e-05,
|
| 1603 |
+
"loss": 0.5251,
|
| 1604 |
+
"step": 215
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 5.684210526315789,
|
| 1608 |
+
"grad_norm": 378.4760437011719,
|
| 1609 |
+
"learning_rate": 2.926829268292683e-05,
|
| 1610 |
+
"loss": 1.1182,
|
| 1611 |
+
"step": 216
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 5.7105263157894735,
|
| 1615 |
+
"grad_norm": 25.041152954101562,
|
| 1616 |
+
"learning_rate": 2.9146341463414634e-05,
|
| 1617 |
+
"loss": 0.5151,
|
| 1618 |
+
"step": 217
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 5.7368421052631575,
|
| 1622 |
+
"grad_norm": 104.20264434814453,
|
| 1623 |
+
"learning_rate": 2.9024390243902438e-05,
|
| 1624 |
+
"loss": 0.8081,
|
| 1625 |
+
"step": 218
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 5.7631578947368425,
|
| 1629 |
+
"grad_norm": 71.08457946777344,
|
| 1630 |
+
"learning_rate": 2.8902439024390242e-05,
|
| 1631 |
+
"loss": 0.993,
|
| 1632 |
+
"step": 219
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 5.7894736842105265,
|
| 1636 |
+
"grad_norm": 64.38380432128906,
|
| 1637 |
+
"learning_rate": 2.8780487804878046e-05,
|
| 1638 |
+
"loss": 0.8127,
|
| 1639 |
+
"step": 220
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 5.815789473684211,
|
| 1643 |
+
"grad_norm": 28.343429565429688,
|
| 1644 |
+
"learning_rate": 2.8658536585365854e-05,
|
| 1645 |
+
"loss": 0.4542,
|
| 1646 |
+
"step": 221
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 5.842105263157895,
|
| 1650 |
+
"grad_norm": 46.55160140991211,
|
| 1651 |
+
"learning_rate": 2.8536585365853658e-05,
|
| 1652 |
+
"loss": 0.5087,
|
| 1653 |
+
"step": 222
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 5.868421052631579,
|
| 1657 |
+
"grad_norm": 111.2691879272461,
|
| 1658 |
+
"learning_rate": 2.8414634146341462e-05,
|
| 1659 |
+
"loss": 0.4867,
|
| 1660 |
+
"step": 223
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 5.894736842105263,
|
| 1664 |
+
"grad_norm": 76.03248596191406,
|
| 1665 |
+
"learning_rate": 2.8292682926829267e-05,
|
| 1666 |
+
"loss": 0.9315,
|
| 1667 |
+
"step": 224
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 5.921052631578947,
|
| 1671 |
+
"grad_norm": 58.543758392333984,
|
| 1672 |
+
"learning_rate": 2.817073170731707e-05,
|
| 1673 |
+
"loss": 0.398,
|
| 1674 |
+
"step": 225
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 5.947368421052632,
|
| 1678 |
+
"grad_norm": 69.87763214111328,
|
| 1679 |
+
"learning_rate": 2.8048780487804882e-05,
|
| 1680 |
+
"loss": 1.0981,
|
| 1681 |
+
"step": 226
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 5.973684210526316,
|
| 1685 |
+
"grad_norm": 107.25030517578125,
|
| 1686 |
+
"learning_rate": 2.7926829268292686e-05,
|
| 1687 |
+
"loss": 0.7306,
|
| 1688 |
+
"step": 227
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 6.0,
|
| 1692 |
+
"grad_norm": 104.7223129272461,
|
| 1693 |
+
"learning_rate": 2.780487804878049e-05,
|
| 1694 |
+
"loss": 0.4531,
|
| 1695 |
+
"step": 228
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 6.0,
|
| 1699 |
+
"eval_accuracy": 0.6333333333333333,
|
| 1700 |
+
"eval_f1_macro": 0.5696302606408418,
|
| 1701 |
+
"eval_f1_micro": 0.6333333333333333,
|
| 1702 |
+
"eval_f1_weighted": 0.5696302606408418,
|
| 1703 |
+
"eval_loss": 1.0229661464691162,
|
| 1704 |
+
"eval_precision_macro": 0.6182503770739065,
|
| 1705 |
+
"eval_precision_micro": 0.6333333333333333,
|
| 1706 |
+
"eval_precision_weighted": 0.6182503770739065,
|
| 1707 |
+
"eval_recall_macro": 0.6333333333333334,
|
| 1708 |
+
"eval_recall_micro": 0.6333333333333333,
|
| 1709 |
+
"eval_recall_weighted": 0.6333333333333333,
|
| 1710 |
+
"eval_runtime": 69.4357,
|
| 1711 |
+
"eval_samples_per_second": 0.864,
|
| 1712 |
+
"eval_steps_per_second": 0.029,
|
| 1713 |
+
"step": 228
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 6.026315789473684,
|
| 1717 |
+
"grad_norm": 49.86199951171875,
|
| 1718 |
+
"learning_rate": 2.7682926829268298e-05,
|
| 1719 |
+
"loss": 0.4233,
|
| 1720 |
+
"step": 229
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 6.052631578947368,
|
| 1724 |
+
"grad_norm": 47.08704376220703,
|
| 1725 |
+
"learning_rate": 2.7560975609756102e-05,
|
| 1726 |
+
"loss": 0.514,
|
| 1727 |
+
"step": 230
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 6.078947368421052,
|
| 1731 |
+
"grad_norm": 80.2113037109375,
|
| 1732 |
+
"learning_rate": 2.7439024390243906e-05,
|
| 1733 |
+
"loss": 0.4019,
|
| 1734 |
+
"step": 231
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 6.105263157894737,
|
| 1738 |
+
"grad_norm": 23.823795318603516,
|
| 1739 |
+
"learning_rate": 2.731707317073171e-05,
|
| 1740 |
+
"loss": 0.2943,
|
| 1741 |
+
"step": 232
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 6.131578947368421,
|
| 1745 |
+
"grad_norm": 299.1929016113281,
|
| 1746 |
+
"learning_rate": 2.7195121951219515e-05,
|
| 1747 |
+
"loss": 0.3543,
|
| 1748 |
+
"step": 233
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 6.157894736842105,
|
| 1752 |
+
"grad_norm": 73.6047592163086,
|
| 1753 |
+
"learning_rate": 2.707317073170732e-05,
|
| 1754 |
+
"loss": 0.4757,
|
| 1755 |
+
"step": 234
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 6.184210526315789,
|
| 1759 |
+
"grad_norm": 30.344865798950195,
|
| 1760 |
+
"learning_rate": 2.6951219512195123e-05,
|
| 1761 |
+
"loss": 0.3526,
|
| 1762 |
+
"step": 235
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 6.2105263157894735,
|
| 1766 |
+
"grad_norm": 98.406494140625,
|
| 1767 |
+
"learning_rate": 2.682926829268293e-05,
|
| 1768 |
+
"loss": 0.8068,
|
| 1769 |
+
"step": 236
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 6.2368421052631575,
|
| 1773 |
+
"grad_norm": 95.38936614990234,
|
| 1774 |
+
"learning_rate": 2.6707317073170735e-05,
|
| 1775 |
+
"loss": 0.2938,
|
| 1776 |
+
"step": 237
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 6.2631578947368425,
|
| 1780 |
+
"grad_norm": 90.7430648803711,
|
| 1781 |
+
"learning_rate": 2.658536585365854e-05,
|
| 1782 |
+
"loss": 0.4397,
|
| 1783 |
+
"step": 238
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 6.2894736842105265,
|
| 1787 |
+
"grad_norm": 64.78533172607422,
|
| 1788 |
+
"learning_rate": 2.6463414634146344e-05,
|
| 1789 |
+
"loss": 0.4477,
|
| 1790 |
+
"step": 239
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 6.315789473684211,
|
| 1794 |
+
"grad_norm": 20.26727294921875,
|
| 1795 |
+
"learning_rate": 2.6341463414634148e-05,
|
| 1796 |
+
"loss": 0.2176,
|
| 1797 |
+
"step": 240
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 6.342105263157895,
|
| 1801 |
+
"grad_norm": 39.570987701416016,
|
| 1802 |
+
"learning_rate": 2.6219512195121952e-05,
|
| 1803 |
+
"loss": 0.4976,
|
| 1804 |
+
"step": 241
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 6.368421052631579,
|
| 1808 |
+
"grad_norm": 56.996559143066406,
|
| 1809 |
+
"learning_rate": 2.609756097560976e-05,
|
| 1810 |
+
"loss": 0.6441,
|
| 1811 |
+
"step": 242
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 6.394736842105263,
|
| 1815 |
+
"grad_norm": 172.96588134765625,
|
| 1816 |
+
"learning_rate": 2.5975609756097564e-05,
|
| 1817 |
+
"loss": 0.3586,
|
| 1818 |
+
"step": 243
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 6.421052631578947,
|
| 1822 |
+
"grad_norm": 39.748512268066406,
|
| 1823 |
+
"learning_rate": 2.5853658536585368e-05,
|
| 1824 |
+
"loss": 0.4,
|
| 1825 |
+
"step": 244
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 6.447368421052632,
|
| 1829 |
+
"grad_norm": 47.228515625,
|
| 1830 |
+
"learning_rate": 2.5731707317073172e-05,
|
| 1831 |
+
"loss": 0.5478,
|
| 1832 |
+
"step": 245
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 6.473684210526316,
|
| 1836 |
+
"grad_norm": 83.61058807373047,
|
| 1837 |
+
"learning_rate": 2.5609756097560977e-05,
|
| 1838 |
+
"loss": 0.6447,
|
| 1839 |
+
"step": 246
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 6.5,
|
| 1843 |
+
"grad_norm": 96.23981475830078,
|
| 1844 |
+
"learning_rate": 2.548780487804878e-05,
|
| 1845 |
+
"loss": 0.6428,
|
| 1846 |
+
"step": 247
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 6.526315789473684,
|
| 1850 |
+
"grad_norm": 91.69573974609375,
|
| 1851 |
+
"learning_rate": 2.536585365853659e-05,
|
| 1852 |
+
"loss": 0.5696,
|
| 1853 |
+
"step": 248
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 6.552631578947368,
|
| 1857 |
+
"grad_norm": 80.09562683105469,
|
| 1858 |
+
"learning_rate": 2.5243902439024393e-05,
|
| 1859 |
+
"loss": 0.5676,
|
| 1860 |
+
"step": 249
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 6.578947368421053,
|
| 1864 |
+
"grad_norm": 71.28556060791016,
|
| 1865 |
+
"learning_rate": 2.5121951219512197e-05,
|
| 1866 |
+
"loss": 0.5665,
|
| 1867 |
+
"step": 250
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 6.605263157894737,
|
| 1871 |
+
"grad_norm": 159.1014862060547,
|
| 1872 |
+
"learning_rate": 2.5e-05,
|
| 1873 |
+
"loss": 0.6344,
|
| 1874 |
+
"step": 251
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 6.631578947368421,
|
| 1878 |
+
"grad_norm": 101.53102111816406,
|
| 1879 |
+
"learning_rate": 2.4878048780487805e-05,
|
| 1880 |
+
"loss": 0.52,
|
| 1881 |
+
"step": 252
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 6.657894736842105,
|
| 1885 |
+
"grad_norm": 80.52534484863281,
|
| 1886 |
+
"learning_rate": 2.475609756097561e-05,
|
| 1887 |
+
"loss": 0.4053,
|
| 1888 |
+
"step": 253
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"epoch": 6.684210526315789,
|
| 1892 |
+
"grad_norm": 66.0854263305664,
|
| 1893 |
+
"learning_rate": 2.4634146341463414e-05,
|
| 1894 |
+
"loss": 0.4303,
|
| 1895 |
+
"step": 254
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 6.7105263157894735,
|
| 1899 |
+
"grad_norm": 97.85746002197266,
|
| 1900 |
+
"learning_rate": 2.451219512195122e-05,
|
| 1901 |
+
"loss": 0.3143,
|
| 1902 |
+
"step": 255
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 6.7368421052631575,
|
| 1906 |
+
"grad_norm": 57.940155029296875,
|
| 1907 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 1908 |
+
"loss": 0.5949,
|
| 1909 |
+
"step": 256
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 6.7631578947368425,
|
| 1913 |
+
"grad_norm": 40.67607879638672,
|
| 1914 |
+
"learning_rate": 2.426829268292683e-05,
|
| 1915 |
+
"loss": 0.4334,
|
| 1916 |
+
"step": 257
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 6.7894736842105265,
|
| 1920 |
+
"grad_norm": 133.9537811279297,
|
| 1921 |
+
"learning_rate": 2.4146341463414634e-05,
|
| 1922 |
+
"loss": 0.3846,
|
| 1923 |
+
"step": 258
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 6.815789473684211,
|
| 1927 |
+
"grad_norm": 83.99858856201172,
|
| 1928 |
+
"learning_rate": 2.402439024390244e-05,
|
| 1929 |
+
"loss": 0.3965,
|
| 1930 |
+
"step": 259
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 6.842105263157895,
|
| 1934 |
+
"grad_norm": 98.94632720947266,
|
| 1935 |
+
"learning_rate": 2.3902439024390243e-05,
|
| 1936 |
+
"loss": 0.6727,
|
| 1937 |
+
"step": 260
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 6.868421052631579,
|
| 1941 |
+
"grad_norm": 121.6048812866211,
|
| 1942 |
+
"learning_rate": 2.378048780487805e-05,
|
| 1943 |
+
"loss": 0.6255,
|
| 1944 |
+
"step": 261
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 6.894736842105263,
|
| 1948 |
+
"grad_norm": 41.528053283691406,
|
| 1949 |
+
"learning_rate": 2.3658536585365854e-05,
|
| 1950 |
+
"loss": 0.5627,
|
| 1951 |
+
"step": 262
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 6.921052631578947,
|
| 1955 |
+
"grad_norm": 95.7408447265625,
|
| 1956 |
+
"learning_rate": 2.353658536585366e-05,
|
| 1957 |
+
"loss": 0.487,
|
| 1958 |
+
"step": 263
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 6.947368421052632,
|
| 1962 |
+
"grad_norm": 48.754642486572266,
|
| 1963 |
+
"learning_rate": 2.3414634146341466e-05,
|
| 1964 |
+
"loss": 0.2175,
|
| 1965 |
+
"step": 264
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 6.973684210526316,
|
| 1969 |
+
"grad_norm": 39.36869430541992,
|
| 1970 |
+
"learning_rate": 2.329268292682927e-05,
|
| 1971 |
+
"loss": 0.3606,
|
| 1972 |
+
"step": 265
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 7.0,
|
| 1976 |
+
"grad_norm": 116.08528900146484,
|
| 1977 |
+
"learning_rate": 2.3170731707317075e-05,
|
| 1978 |
+
"loss": 0.4935,
|
| 1979 |
+
"step": 266
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 7.0,
|
| 1983 |
+
"eval_accuracy": 0.6333333333333333,
|
| 1984 |
+
"eval_f1_macro": 0.6192316540142626,
|
| 1985 |
+
"eval_f1_micro": 0.6333333333333333,
|
| 1986 |
+
"eval_f1_weighted": 0.6192316540142627,
|
| 1987 |
+
"eval_loss": 0.8341596126556396,
|
| 1988 |
+
"eval_precision_macro": 0.6416666666666667,
|
| 1989 |
+
"eval_precision_micro": 0.6333333333333333,
|
| 1990 |
+
"eval_precision_weighted": 0.6416666666666667,
|
| 1991 |
+
"eval_recall_macro": 0.6333333333333333,
|
| 1992 |
+
"eval_recall_micro": 0.6333333333333333,
|
| 1993 |
+
"eval_recall_weighted": 0.6333333333333333,
|
| 1994 |
+
"eval_runtime": 67.0822,
|
| 1995 |
+
"eval_samples_per_second": 0.894,
|
| 1996 |
+
"eval_steps_per_second": 0.03,
|
| 1997 |
+
"step": 266
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 7.026315789473684,
|
| 2001 |
+
"grad_norm": 33.8989143371582,
|
| 2002 |
+
"learning_rate": 2.304878048780488e-05,
|
| 2003 |
+
"loss": 0.3444,
|
| 2004 |
+
"step": 267
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 7.052631578947368,
|
| 2008 |
+
"grad_norm": 37.1051139831543,
|
| 2009 |
+
"learning_rate": 2.2926829268292687e-05,
|
| 2010 |
+
"loss": 0.4544,
|
| 2011 |
+
"step": 268
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 7.078947368421052,
|
| 2015 |
+
"grad_norm": 65.14508056640625,
|
| 2016 |
+
"learning_rate": 2.280487804878049e-05,
|
| 2017 |
+
"loss": 0.6778,
|
| 2018 |
+
"step": 269
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 7.105263157894737,
|
| 2022 |
+
"grad_norm": 64.3533706665039,
|
| 2023 |
+
"learning_rate": 2.2682926829268295e-05,
|
| 2024 |
+
"loss": 0.4157,
|
| 2025 |
+
"step": 270
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 7.131578947368421,
|
| 2029 |
+
"grad_norm": 26.499357223510742,
|
| 2030 |
+
"learning_rate": 2.25609756097561e-05,
|
| 2031 |
+
"loss": 0.243,
|
| 2032 |
+
"step": 271
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 7.157894736842105,
|
| 2036 |
+
"grad_norm": 48.940486907958984,
|
| 2037 |
+
"learning_rate": 2.2439024390243904e-05,
|
| 2038 |
+
"loss": 0.3446,
|
| 2039 |
+
"step": 272
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 7.184210526315789,
|
| 2043 |
+
"grad_norm": 31.625215530395508,
|
| 2044 |
+
"learning_rate": 2.2317073170731708e-05,
|
| 2045 |
+
"loss": 0.2034,
|
| 2046 |
+
"step": 273
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 7.2105263157894735,
|
| 2050 |
+
"grad_norm": 25.036733627319336,
|
| 2051 |
+
"learning_rate": 2.2195121951219512e-05,
|
| 2052 |
+
"loss": 0.1459,
|
| 2053 |
+
"step": 274
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 7.2368421052631575,
|
| 2057 |
+
"grad_norm": 81.56002807617188,
|
| 2058 |
+
"learning_rate": 2.207317073170732e-05,
|
| 2059 |
+
"loss": 0.4705,
|
| 2060 |
+
"step": 275
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 7.2631578947368425,
|
| 2064 |
+
"grad_norm": 97.85039520263672,
|
| 2065 |
+
"learning_rate": 2.1951219512195124e-05,
|
| 2066 |
+
"loss": 0.3849,
|
| 2067 |
+
"step": 276
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 7.2894736842105265,
|
| 2071 |
+
"grad_norm": 92.32601165771484,
|
| 2072 |
+
"learning_rate": 2.1829268292682928e-05,
|
| 2073 |
+
"loss": 0.4016,
|
| 2074 |
+
"step": 277
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 7.315789473684211,
|
| 2078 |
+
"grad_norm": 52.31543731689453,
|
| 2079 |
+
"learning_rate": 2.1707317073170732e-05,
|
| 2080 |
+
"loss": 0.2492,
|
| 2081 |
+
"step": 278
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 7.342105263157895,
|
| 2085 |
+
"grad_norm": 103.44355010986328,
|
| 2086 |
+
"learning_rate": 2.1585365853658537e-05,
|
| 2087 |
+
"loss": 0.4627,
|
| 2088 |
+
"step": 279
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 7.368421052631579,
|
| 2092 |
+
"grad_norm": 124.98883056640625,
|
| 2093 |
+
"learning_rate": 2.146341463414634e-05,
|
| 2094 |
+
"loss": 0.4096,
|
| 2095 |
+
"step": 280
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 7.394736842105263,
|
| 2099 |
+
"grad_norm": 58.3911018371582,
|
| 2100 |
+
"learning_rate": 2.134146341463415e-05,
|
| 2101 |
+
"loss": 0.1543,
|
| 2102 |
+
"step": 281
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 7.421052631578947,
|
| 2106 |
+
"grad_norm": 65.31929016113281,
|
| 2107 |
+
"learning_rate": 2.1219512195121953e-05,
|
| 2108 |
+
"loss": 0.5474,
|
| 2109 |
+
"step": 282
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 7.447368421052632,
|
| 2113 |
+
"grad_norm": 22.333057403564453,
|
| 2114 |
+
"learning_rate": 2.1097560975609757e-05,
|
| 2115 |
+
"loss": 0.2357,
|
| 2116 |
+
"step": 283
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 7.473684210526316,
|
| 2120 |
+
"grad_norm": 43.33408737182617,
|
| 2121 |
+
"learning_rate": 2.097560975609756e-05,
|
| 2122 |
+
"loss": 0.1541,
|
| 2123 |
+
"step": 284
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 7.5,
|
| 2127 |
+
"grad_norm": 107.57859802246094,
|
| 2128 |
+
"learning_rate": 2.0853658536585365e-05,
|
| 2129 |
+
"loss": 0.239,
|
| 2130 |
+
"step": 285
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 7.526315789473684,
|
| 2134 |
+
"grad_norm": 69.31283569335938,
|
| 2135 |
+
"learning_rate": 2.073170731707317e-05,
|
| 2136 |
+
"loss": 0.5448,
|
| 2137 |
+
"step": 286
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 7.552631578947368,
|
| 2141 |
+
"grad_norm": 102.60783386230469,
|
| 2142 |
+
"learning_rate": 2.0609756097560977e-05,
|
| 2143 |
+
"loss": 0.3203,
|
| 2144 |
+
"step": 287
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 7.578947368421053,
|
| 2148 |
+
"grad_norm": 112.20484924316406,
|
| 2149 |
+
"learning_rate": 2.048780487804878e-05,
|
| 2150 |
+
"loss": 0.4882,
|
| 2151 |
+
"step": 288
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 7.605263157894737,
|
| 2155 |
+
"grad_norm": 69.58258819580078,
|
| 2156 |
+
"learning_rate": 2.0365853658536586e-05,
|
| 2157 |
+
"loss": 0.8047,
|
| 2158 |
+
"step": 289
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 7.631578947368421,
|
| 2162 |
+
"grad_norm": 37.90599822998047,
|
| 2163 |
+
"learning_rate": 2.0243902439024393e-05,
|
| 2164 |
+
"loss": 0.1967,
|
| 2165 |
+
"step": 290
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 7.657894736842105,
|
| 2169 |
+
"grad_norm": 40.24911880493164,
|
| 2170 |
+
"learning_rate": 2.0121951219512197e-05,
|
| 2171 |
+
"loss": 0.4404,
|
| 2172 |
+
"step": 291
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 7.684210526315789,
|
| 2176 |
+
"grad_norm": 70.96076202392578,
|
| 2177 |
+
"learning_rate": 2e-05,
|
| 2178 |
+
"loss": 0.2328,
|
| 2179 |
+
"step": 292
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 7.7105263157894735,
|
| 2183 |
+
"grad_norm": 60.926185607910156,
|
| 2184 |
+
"learning_rate": 1.9878048780487806e-05,
|
| 2185 |
+
"loss": 0.4773,
|
| 2186 |
+
"step": 293
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 7.7368421052631575,
|
| 2190 |
+
"grad_norm": 25.279253005981445,
|
| 2191 |
+
"learning_rate": 1.975609756097561e-05,
|
| 2192 |
+
"loss": 0.3064,
|
| 2193 |
+
"step": 294
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 7.7631578947368425,
|
| 2197 |
+
"grad_norm": 188.24697875976562,
|
| 2198 |
+
"learning_rate": 1.9634146341463418e-05,
|
| 2199 |
+
"loss": 0.3841,
|
| 2200 |
+
"step": 295
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 7.7894736842105265,
|
| 2204 |
+
"grad_norm": 78.81002807617188,
|
| 2205 |
+
"learning_rate": 1.9512195121951222e-05,
|
| 2206 |
+
"loss": 0.243,
|
| 2207 |
+
"step": 296
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 7.815789473684211,
|
| 2211 |
+
"grad_norm": 50.173526763916016,
|
| 2212 |
+
"learning_rate": 1.9390243902439026e-05,
|
| 2213 |
+
"loss": 0.4069,
|
| 2214 |
+
"step": 297
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 7.842105263157895,
|
| 2218 |
+
"grad_norm": 16.644210815429688,
|
| 2219 |
+
"learning_rate": 1.926829268292683e-05,
|
| 2220 |
+
"loss": 0.1207,
|
| 2221 |
+
"step": 298
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 7.868421052631579,
|
| 2225 |
+
"grad_norm": 318.4234924316406,
|
| 2226 |
+
"learning_rate": 1.9146341463414635e-05,
|
| 2227 |
+
"loss": 0.7463,
|
| 2228 |
+
"step": 299
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 7.894736842105263,
|
| 2232 |
+
"grad_norm": 61.193946838378906,
|
| 2233 |
+
"learning_rate": 1.902439024390244e-05,
|
| 2234 |
+
"loss": 0.4838,
|
| 2235 |
+
"step": 300
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 7.921052631578947,
|
| 2239 |
+
"grad_norm": 42.694698333740234,
|
| 2240 |
+
"learning_rate": 1.8902439024390246e-05,
|
| 2241 |
+
"loss": 0.2745,
|
| 2242 |
+
"step": 301
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 7.947368421052632,
|
| 2246 |
+
"grad_norm": 97.97174835205078,
|
| 2247 |
+
"learning_rate": 1.878048780487805e-05,
|
| 2248 |
+
"loss": 0.6297,
|
| 2249 |
+
"step": 302
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 7.973684210526316,
|
| 2253 |
+
"grad_norm": 98.98281860351562,
|
| 2254 |
+
"learning_rate": 1.8658536585365855e-05,
|
| 2255 |
+
"loss": 0.5306,
|
| 2256 |
+
"step": 303
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 8.0,
|
| 2260 |
+
"grad_norm": 82.52599334716797,
|
| 2261 |
+
"learning_rate": 1.853658536585366e-05,
|
| 2262 |
+
"loss": 0.0833,
|
| 2263 |
+
"step": 304
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 8.0,
|
| 2267 |
+
"eval_accuracy": 0.6333333333333333,
|
| 2268 |
+
"eval_f1_macro": 0.6299972429004687,
|
| 2269 |
+
"eval_f1_micro": 0.6333333333333333,
|
| 2270 |
+
"eval_f1_weighted": 0.6299972429004687,
|
| 2271 |
+
"eval_loss": 0.8269728422164917,
|
| 2272 |
+
"eval_precision_macro": 0.6510416666666666,
|
| 2273 |
+
"eval_precision_micro": 0.6333333333333333,
|
| 2274 |
+
"eval_precision_weighted": 0.6510416666666666,
|
| 2275 |
+
"eval_recall_macro": 0.6333333333333333,
|
| 2276 |
+
"eval_recall_micro": 0.6333333333333333,
|
| 2277 |
+
"eval_recall_weighted": 0.6333333333333333,
|
| 2278 |
+
"eval_runtime": 71.9256,
|
| 2279 |
+
"eval_samples_per_second": 0.834,
|
| 2280 |
+
"eval_steps_per_second": 0.028,
|
| 2281 |
+
"step": 304
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 8.026315789473685,
|
| 2285 |
+
"grad_norm": 21.338314056396484,
|
| 2286 |
+
"learning_rate": 1.8414634146341463e-05,
|
| 2287 |
+
"loss": 0.1345,
|
| 2288 |
+
"step": 305
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 8.052631578947368,
|
| 2292 |
+
"grad_norm": 46.4133186340332,
|
| 2293 |
+
"learning_rate": 1.8292682926829268e-05,
|
| 2294 |
+
"loss": 0.225,
|
| 2295 |
+
"step": 306
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 8.078947368421053,
|
| 2299 |
+
"grad_norm": 65.10118865966797,
|
| 2300 |
+
"learning_rate": 1.8170731707317075e-05,
|
| 2301 |
+
"loss": 0.2522,
|
| 2302 |
+
"step": 307
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 8.105263157894736,
|
| 2306 |
+
"grad_norm": 55.527183532714844,
|
| 2307 |
+
"learning_rate": 1.804878048780488e-05,
|
| 2308 |
+
"loss": 0.3192,
|
| 2309 |
+
"step": 308
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 8.131578947368421,
|
| 2313 |
+
"grad_norm": 185.77838134765625,
|
| 2314 |
+
"learning_rate": 1.7926829268292684e-05,
|
| 2315 |
+
"loss": 0.3924,
|
| 2316 |
+
"step": 309
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 8.157894736842104,
|
| 2320 |
+
"grad_norm": 37.39326095581055,
|
| 2321 |
+
"learning_rate": 1.7804878048780488e-05,
|
| 2322 |
+
"loss": 0.3674,
|
| 2323 |
+
"step": 310
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 8.18421052631579,
|
| 2327 |
+
"grad_norm": 120.3980941772461,
|
| 2328 |
+
"learning_rate": 1.7682926829268292e-05,
|
| 2329 |
+
"loss": 0.3816,
|
| 2330 |
+
"step": 311
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 8.210526315789474,
|
| 2334 |
+
"grad_norm": 114.17828369140625,
|
| 2335 |
+
"learning_rate": 1.7560975609756096e-05,
|
| 2336 |
+
"loss": 0.5185,
|
| 2337 |
+
"step": 312
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 8.236842105263158,
|
| 2341 |
+
"grad_norm": 216.08956909179688,
|
| 2342 |
+
"learning_rate": 1.74390243902439e-05,
|
| 2343 |
+
"loss": 0.3604,
|
| 2344 |
+
"step": 313
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 8.263157894736842,
|
| 2348 |
+
"grad_norm": 83.65510559082031,
|
| 2349 |
+
"learning_rate": 1.7317073170731708e-05,
|
| 2350 |
+
"loss": 0.2593,
|
| 2351 |
+
"step": 314
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 8.289473684210526,
|
| 2355 |
+
"grad_norm": 49.834617614746094,
|
| 2356 |
+
"learning_rate": 1.7195121951219512e-05,
|
| 2357 |
+
"loss": 0.2843,
|
| 2358 |
+
"step": 315
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 8.31578947368421,
|
| 2362 |
+
"grad_norm": 130.27406311035156,
|
| 2363 |
+
"learning_rate": 1.707317073170732e-05,
|
| 2364 |
+
"loss": 0.4344,
|
| 2365 |
+
"step": 316
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 8.342105263157896,
|
| 2369 |
+
"grad_norm": 218.7544403076172,
|
| 2370 |
+
"learning_rate": 1.6951219512195124e-05,
|
| 2371 |
+
"loss": 0.7271,
|
| 2372 |
+
"step": 317
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 8.368421052631579,
|
| 2376 |
+
"grad_norm": 54.966861724853516,
|
| 2377 |
+
"learning_rate": 1.682926829268293e-05,
|
| 2378 |
+
"loss": 0.2361,
|
| 2379 |
+
"step": 318
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 8.394736842105264,
|
| 2383 |
+
"grad_norm": 33.67280197143555,
|
| 2384 |
+
"learning_rate": 1.6707317073170733e-05,
|
| 2385 |
+
"loss": 0.2209,
|
| 2386 |
+
"step": 319
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 8.421052631578947,
|
| 2390 |
+
"grad_norm": 50.271949768066406,
|
| 2391 |
+
"learning_rate": 1.6585365853658537e-05,
|
| 2392 |
+
"loss": 0.3898,
|
| 2393 |
+
"step": 320
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 8.447368421052632,
|
| 2397 |
+
"grad_norm": 43.207115173339844,
|
| 2398 |
+
"learning_rate": 1.6463414634146345e-05,
|
| 2399 |
+
"loss": 0.3961,
|
| 2400 |
+
"step": 321
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 8.473684210526315,
|
| 2404 |
+
"grad_norm": 33.84910583496094,
|
| 2405 |
+
"learning_rate": 1.634146341463415e-05,
|
| 2406 |
+
"loss": 0.146,
|
| 2407 |
+
"step": 322
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 8.5,
|
| 2411 |
+
"grad_norm": 22.738853454589844,
|
| 2412 |
+
"learning_rate": 1.6219512195121953e-05,
|
| 2413 |
+
"loss": 0.1157,
|
| 2414 |
+
"step": 323
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 8.526315789473685,
|
| 2418 |
+
"grad_norm": 25.66840362548828,
|
| 2419 |
+
"learning_rate": 1.6097560975609757e-05,
|
| 2420 |
+
"loss": 0.218,
|
| 2421 |
+
"step": 324
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 8.552631578947368,
|
| 2425 |
+
"grad_norm": 7.81503438949585,
|
| 2426 |
+
"learning_rate": 1.597560975609756e-05,
|
| 2427 |
+
"loss": 0.0832,
|
| 2428 |
+
"step": 325
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 8.578947368421053,
|
| 2432 |
+
"grad_norm": 40.258338928222656,
|
| 2433 |
+
"learning_rate": 1.5853658536585366e-05,
|
| 2434 |
+
"loss": 0.1769,
|
| 2435 |
+
"step": 326
|
| 2436 |
+
},
|
| 2437 |
+
{
|
| 2438 |
+
"epoch": 8.605263157894736,
|
| 2439 |
+
"grad_norm": 70.59412384033203,
|
| 2440 |
+
"learning_rate": 1.573170731707317e-05,
|
| 2441 |
+
"loss": 0.4184,
|
| 2442 |
+
"step": 327
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 8.631578947368421,
|
| 2446 |
+
"grad_norm": 52.296180725097656,
|
| 2447 |
+
"learning_rate": 1.5609756097560978e-05,
|
| 2448 |
+
"loss": 0.215,
|
| 2449 |
+
"step": 328
|
| 2450 |
+
},
|
| 2451 |
+
{
|
| 2452 |
+
"epoch": 8.657894736842106,
|
| 2453 |
+
"grad_norm": 100.57954406738281,
|
| 2454 |
+
"learning_rate": 1.5487804878048782e-05,
|
| 2455 |
+
"loss": 0.1613,
|
| 2456 |
+
"step": 329
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 8.68421052631579,
|
| 2460 |
+
"grad_norm": 41.98234558105469,
|
| 2461 |
+
"learning_rate": 1.5365853658536586e-05,
|
| 2462 |
+
"loss": 0.3339,
|
| 2463 |
+
"step": 330
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 8.710526315789474,
|
| 2467 |
+
"grad_norm": 43.48909378051758,
|
| 2468 |
+
"learning_rate": 1.524390243902439e-05,
|
| 2469 |
+
"loss": 0.2052,
|
| 2470 |
+
"step": 331
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 8.736842105263158,
|
| 2474 |
+
"grad_norm": 108.465087890625,
|
| 2475 |
+
"learning_rate": 1.5121951219512196e-05,
|
| 2476 |
+
"loss": 0.2712,
|
| 2477 |
+
"step": 332
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 8.763157894736842,
|
| 2481 |
+
"grad_norm": 35.05289840698242,
|
| 2482 |
+
"learning_rate": 1.5e-05,
|
| 2483 |
+
"loss": 0.0903,
|
| 2484 |
+
"step": 333
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 8.789473684210526,
|
| 2488 |
+
"grad_norm": 21.532867431640625,
|
| 2489 |
+
"learning_rate": 1.4878048780487805e-05,
|
| 2490 |
+
"loss": 0.1428,
|
| 2491 |
+
"step": 334
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 8.81578947368421,
|
| 2495 |
+
"grad_norm": 62.48358154296875,
|
| 2496 |
+
"learning_rate": 1.475609756097561e-05,
|
| 2497 |
+
"loss": 0.242,
|
| 2498 |
+
"step": 335
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 8.842105263157894,
|
| 2502 |
+
"grad_norm": 33.35801696777344,
|
| 2503 |
+
"learning_rate": 1.4634146341463415e-05,
|
| 2504 |
+
"loss": 0.3042,
|
| 2505 |
+
"step": 336
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 8.868421052631579,
|
| 2509 |
+
"grad_norm": 38.024749755859375,
|
| 2510 |
+
"learning_rate": 1.4512195121951219e-05,
|
| 2511 |
+
"loss": 0.2156,
|
| 2512 |
+
"step": 337
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 8.894736842105264,
|
| 2516 |
+
"grad_norm": 34.33394241333008,
|
| 2517 |
+
"learning_rate": 1.4390243902439023e-05,
|
| 2518 |
+
"loss": 0.1868,
|
| 2519 |
+
"step": 338
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 8.921052631578947,
|
| 2523 |
+
"grad_norm": 68.10643768310547,
|
| 2524 |
+
"learning_rate": 1.4268292682926829e-05,
|
| 2525 |
+
"loss": 0.3916,
|
| 2526 |
+
"step": 339
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 8.947368421052632,
|
| 2530 |
+
"grad_norm": 108.21729278564453,
|
| 2531 |
+
"learning_rate": 1.4146341463414633e-05,
|
| 2532 |
+
"loss": 0.3538,
|
| 2533 |
+
"step": 340
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 8.973684210526315,
|
| 2537 |
+
"grad_norm": 39.021934509277344,
|
| 2538 |
+
"learning_rate": 1.4024390243902441e-05,
|
| 2539 |
+
"loss": 0.2256,
|
| 2540 |
+
"step": 341
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 9.0,
|
| 2544 |
+
"grad_norm": 62.53577423095703,
|
| 2545 |
+
"learning_rate": 1.3902439024390245e-05,
|
| 2546 |
+
"loss": 0.1077,
|
| 2547 |
+
"step": 342
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 9.0,
|
| 2551 |
+
"eval_accuracy": 0.65,
|
| 2552 |
+
"eval_f1_macro": 0.6464173059000645,
|
| 2553 |
+
"eval_f1_micro": 0.65,
|
| 2554 |
+
"eval_f1_weighted": 0.6464173059000645,
|
| 2555 |
+
"eval_loss": 0.7723664045333862,
|
| 2556 |
+
"eval_precision_macro": 0.6687062937062936,
|
| 2557 |
+
"eval_precision_micro": 0.65,
|
| 2558 |
+
"eval_precision_weighted": 0.6687062937062938,
|
| 2559 |
+
"eval_recall_macro": 0.65,
|
| 2560 |
+
"eval_recall_micro": 0.65,
|
| 2561 |
+
"eval_recall_weighted": 0.65,
|
| 2562 |
+
"eval_runtime": 81.5249,
|
| 2563 |
+
"eval_samples_per_second": 0.736,
|
| 2564 |
+
"eval_steps_per_second": 0.025,
|
| 2565 |
+
"step": 342
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 9.026315789473685,
|
| 2569 |
+
"grad_norm": 44.54621124267578,
|
| 2570 |
+
"learning_rate": 1.3780487804878051e-05,
|
| 2571 |
+
"loss": 0.1634,
|
| 2572 |
+
"step": 343
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 9.052631578947368,
|
| 2576 |
+
"grad_norm": 51.46065139770508,
|
| 2577 |
+
"learning_rate": 1.3658536585365855e-05,
|
| 2578 |
+
"loss": 0.118,
|
| 2579 |
+
"step": 344
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 9.078947368421053,
|
| 2583 |
+
"grad_norm": 46.26784133911133,
|
| 2584 |
+
"learning_rate": 1.353658536585366e-05,
|
| 2585 |
+
"loss": 0.2486,
|
| 2586 |
+
"step": 345
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 9.105263157894736,
|
| 2590 |
+
"grad_norm": 43.58497619628906,
|
| 2591 |
+
"learning_rate": 1.3414634146341466e-05,
|
| 2592 |
+
"loss": 0.2445,
|
| 2593 |
+
"step": 346
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 9.131578947368421,
|
| 2597 |
+
"grad_norm": 40.656192779541016,
|
| 2598 |
+
"learning_rate": 1.329268292682927e-05,
|
| 2599 |
+
"loss": 0.1004,
|
| 2600 |
+
"step": 347
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 9.157894736842104,
|
| 2604 |
+
"grad_norm": 55.23797607421875,
|
| 2605 |
+
"learning_rate": 1.3170731707317074e-05,
|
| 2606 |
+
"loss": 0.1722,
|
| 2607 |
+
"step": 348
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 9.18421052631579,
|
| 2611 |
+
"grad_norm": 52.0846061706543,
|
| 2612 |
+
"learning_rate": 1.304878048780488e-05,
|
| 2613 |
+
"loss": 0.1975,
|
| 2614 |
+
"step": 349
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 9.210526315789474,
|
| 2618 |
+
"grad_norm": 38.67387390136719,
|
| 2619 |
+
"learning_rate": 1.2926829268292684e-05,
|
| 2620 |
+
"loss": 0.1192,
|
| 2621 |
+
"step": 350
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 9.236842105263158,
|
| 2625 |
+
"grad_norm": 29.285099029541016,
|
| 2626 |
+
"learning_rate": 1.2804878048780488e-05,
|
| 2627 |
+
"loss": 0.1389,
|
| 2628 |
+
"step": 351
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 9.263157894736842,
|
| 2632 |
+
"grad_norm": 22.35875129699707,
|
| 2633 |
+
"learning_rate": 1.2682926829268294e-05,
|
| 2634 |
+
"loss": 0.0819,
|
| 2635 |
+
"step": 352
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 9.289473684210526,
|
| 2639 |
+
"grad_norm": 12.087577819824219,
|
| 2640 |
+
"learning_rate": 1.2560975609756098e-05,
|
| 2641 |
+
"loss": 0.0583,
|
| 2642 |
+
"step": 353
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 9.31578947368421,
|
| 2646 |
+
"grad_norm": 17.734233856201172,
|
| 2647 |
+
"learning_rate": 1.2439024390243903e-05,
|
| 2648 |
+
"loss": 0.1236,
|
| 2649 |
+
"step": 354
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 9.342105263157896,
|
| 2653 |
+
"grad_norm": 22.403324127197266,
|
| 2654 |
+
"learning_rate": 1.2317073170731707e-05,
|
| 2655 |
+
"loss": 0.1625,
|
| 2656 |
+
"step": 355
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 9.368421052631579,
|
| 2660 |
+
"grad_norm": 7.183572292327881,
|
| 2661 |
+
"learning_rate": 1.2195121951219513e-05,
|
| 2662 |
+
"loss": 0.053,
|
| 2663 |
+
"step": 356
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 9.394736842105264,
|
| 2667 |
+
"grad_norm": 79.79871368408203,
|
| 2668 |
+
"learning_rate": 1.2073170731707317e-05,
|
| 2669 |
+
"loss": 0.3546,
|
| 2670 |
+
"step": 357
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 9.421052631578947,
|
| 2674 |
+
"grad_norm": 35.34999084472656,
|
| 2675 |
+
"learning_rate": 1.1951219512195121e-05,
|
| 2676 |
+
"loss": 0.2396,
|
| 2677 |
+
"step": 358
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 9.447368421052632,
|
| 2681 |
+
"grad_norm": 43.688331604003906,
|
| 2682 |
+
"learning_rate": 1.1829268292682927e-05,
|
| 2683 |
+
"loss": 0.1568,
|
| 2684 |
+
"step": 359
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 9.473684210526315,
|
| 2688 |
+
"grad_norm": 34.7044792175293,
|
| 2689 |
+
"learning_rate": 1.1707317073170733e-05,
|
| 2690 |
+
"loss": 0.2334,
|
| 2691 |
+
"step": 360
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 9.5,
|
| 2695 |
+
"grad_norm": 22.405776977539062,
|
| 2696 |
+
"learning_rate": 1.1585365853658537e-05,
|
| 2697 |
+
"loss": 0.1796,
|
| 2698 |
+
"step": 361
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 9.526315789473685,
|
| 2702 |
+
"grad_norm": 51.72984313964844,
|
| 2703 |
+
"learning_rate": 1.1463414634146343e-05,
|
| 2704 |
+
"loss": 0.2123,
|
| 2705 |
+
"step": 362
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 9.552631578947368,
|
| 2709 |
+
"grad_norm": 17.692136764526367,
|
| 2710 |
+
"learning_rate": 1.1341463414634148e-05,
|
| 2711 |
+
"loss": 0.0802,
|
| 2712 |
+
"step": 363
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 9.578947368421053,
|
| 2716 |
+
"grad_norm": 48.475830078125,
|
| 2717 |
+
"learning_rate": 1.1219512195121952e-05,
|
| 2718 |
+
"loss": 0.169,
|
| 2719 |
+
"step": 364
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 9.605263157894736,
|
| 2723 |
+
"grad_norm": 29.798110961914062,
|
| 2724 |
+
"learning_rate": 1.1097560975609756e-05,
|
| 2725 |
+
"loss": 0.2492,
|
| 2726 |
+
"step": 365
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 9.631578947368421,
|
| 2730 |
+
"grad_norm": 29.195980072021484,
|
| 2731 |
+
"learning_rate": 1.0975609756097562e-05,
|
| 2732 |
+
"loss": 0.1442,
|
| 2733 |
+
"step": 366
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 9.657894736842106,
|
| 2737 |
+
"grad_norm": 50.09147262573242,
|
| 2738 |
+
"learning_rate": 1.0853658536585366e-05,
|
| 2739 |
+
"loss": 0.1571,
|
| 2740 |
+
"step": 367
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 9.68421052631579,
|
| 2744 |
+
"grad_norm": 121.38156127929688,
|
| 2745 |
+
"learning_rate": 1.073170731707317e-05,
|
| 2746 |
+
"loss": 0.3714,
|
| 2747 |
+
"step": 368
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 9.710526315789474,
|
| 2751 |
+
"grad_norm": 16.941436767578125,
|
| 2752 |
+
"learning_rate": 1.0609756097560976e-05,
|
| 2753 |
+
"loss": 0.1157,
|
| 2754 |
+
"step": 369
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 9.736842105263158,
|
| 2758 |
+
"grad_norm": 59.37800216674805,
|
| 2759 |
+
"learning_rate": 1.048780487804878e-05,
|
| 2760 |
+
"loss": 0.2167,
|
| 2761 |
+
"step": 370
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 9.763157894736842,
|
| 2765 |
+
"grad_norm": 27.14200782775879,
|
| 2766 |
+
"learning_rate": 1.0365853658536585e-05,
|
| 2767 |
+
"loss": 0.1044,
|
| 2768 |
+
"step": 371
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 9.789473684210526,
|
| 2772 |
+
"grad_norm": 65.77669525146484,
|
| 2773 |
+
"learning_rate": 1.024390243902439e-05,
|
| 2774 |
+
"loss": 0.282,
|
| 2775 |
+
"step": 372
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 9.81578947368421,
|
| 2779 |
+
"grad_norm": 38.659385681152344,
|
| 2780 |
+
"learning_rate": 1.0121951219512197e-05,
|
| 2781 |
+
"loss": 0.1081,
|
| 2782 |
+
"step": 373
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 9.842105263157894,
|
| 2786 |
+
"grad_norm": 54.30513381958008,
|
| 2787 |
+
"learning_rate": 1e-05,
|
| 2788 |
+
"loss": 0.3475,
|
| 2789 |
+
"step": 374
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 9.868421052631579,
|
| 2793 |
+
"grad_norm": 25.771577835083008,
|
| 2794 |
+
"learning_rate": 9.878048780487805e-06,
|
| 2795 |
+
"loss": 0.111,
|
| 2796 |
+
"step": 375
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 9.894736842105264,
|
| 2800 |
+
"grad_norm": 9.227066040039062,
|
| 2801 |
+
"learning_rate": 9.756097560975611e-06,
|
| 2802 |
+
"loss": 0.066,
|
| 2803 |
+
"step": 376
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 9.921052631578947,
|
| 2807 |
+
"grad_norm": 84.51748657226562,
|
| 2808 |
+
"learning_rate": 9.634146341463415e-06,
|
| 2809 |
+
"loss": 0.4513,
|
| 2810 |
+
"step": 377
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 9.947368421052632,
|
| 2814 |
+
"grad_norm": 195.07611083984375,
|
| 2815 |
+
"learning_rate": 9.51219512195122e-06,
|
| 2816 |
+
"loss": 0.3559,
|
| 2817 |
+
"step": 378
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 9.973684210526315,
|
| 2821 |
+
"grad_norm": 44.909061431884766,
|
| 2822 |
+
"learning_rate": 9.390243902439025e-06,
|
| 2823 |
+
"loss": 0.3273,
|
| 2824 |
+
"step": 379
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 10.0,
|
| 2828 |
+
"grad_norm": 6.2627434730529785,
|
| 2829 |
+
"learning_rate": 9.26829268292683e-06,
|
| 2830 |
+
"loss": 0.0612,
|
| 2831 |
+
"step": 380
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 10.0,
|
| 2835 |
+
"eval_accuracy": 0.7,
|
| 2836 |
+
"eval_f1_macro": 0.6867794795783927,
|
| 2837 |
+
"eval_f1_micro": 0.7,
|
| 2838 |
+
"eval_f1_weighted": 0.6867794795783927,
|
| 2839 |
+
"eval_loss": 0.7868326306343079,
|
| 2840 |
+
"eval_precision_macro": 0.7062908496732027,
|
| 2841 |
+
"eval_precision_micro": 0.7,
|
| 2842 |
+
"eval_precision_weighted": 0.7062908496732025,
|
| 2843 |
+
"eval_recall_macro": 0.7,
|
| 2844 |
+
"eval_recall_micro": 0.7,
|
| 2845 |
+
"eval_recall_weighted": 0.7,
|
| 2846 |
+
"eval_runtime": 70.9939,
|
| 2847 |
+
"eval_samples_per_second": 0.845,
|
| 2848 |
+
"eval_steps_per_second": 0.028,
|
| 2849 |
+
"step": 380
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 10.026315789473685,
|
| 2853 |
+
"grad_norm": 49.423892974853516,
|
| 2854 |
+
"learning_rate": 9.146341463414634e-06,
|
| 2855 |
+
"loss": 0.144,
|
| 2856 |
+
"step": 381
|
| 2857 |
+
},
|
| 2858 |
+
{
|
| 2859 |
+
"epoch": 10.052631578947368,
|
| 2860 |
+
"grad_norm": 57.79946517944336,
|
| 2861 |
+
"learning_rate": 9.02439024390244e-06,
|
| 2862 |
+
"loss": 0.1253,
|
| 2863 |
+
"step": 382
|
| 2864 |
+
},
|
| 2865 |
+
{
|
| 2866 |
+
"epoch": 10.078947368421053,
|
| 2867 |
+
"grad_norm": 37.29555892944336,
|
| 2868 |
+
"learning_rate": 8.902439024390244e-06,
|
| 2869 |
+
"loss": 0.1052,
|
| 2870 |
+
"step": 383
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 10.105263157894736,
|
| 2874 |
+
"grad_norm": 17.90376091003418,
|
| 2875 |
+
"learning_rate": 8.780487804878048e-06,
|
| 2876 |
+
"loss": 0.0677,
|
| 2877 |
+
"step": 384
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 10.131578947368421,
|
| 2881 |
+
"grad_norm": 19.415559768676758,
|
| 2882 |
+
"learning_rate": 8.658536585365854e-06,
|
| 2883 |
+
"loss": 0.1531,
|
| 2884 |
+
"step": 385
|
| 2885 |
+
},
|
| 2886 |
+
{
|
| 2887 |
+
"epoch": 10.157894736842104,
|
| 2888 |
+
"grad_norm": 30.255104064941406,
|
| 2889 |
+
"learning_rate": 8.53658536585366e-06,
|
| 2890 |
+
"loss": 0.106,
|
| 2891 |
+
"step": 386
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 10.18421052631579,
|
| 2895 |
+
"grad_norm": 7.881728172302246,
|
| 2896 |
+
"learning_rate": 8.414634146341464e-06,
|
| 2897 |
+
"loss": 0.0707,
|
| 2898 |
+
"step": 387
|
| 2899 |
+
},
|
| 2900 |
+
{
|
| 2901 |
+
"epoch": 10.210526315789474,
|
| 2902 |
+
"grad_norm": 73.7071533203125,
|
| 2903 |
+
"learning_rate": 8.292682926829268e-06,
|
| 2904 |
+
"loss": 0.4019,
|
| 2905 |
+
"step": 388
|
| 2906 |
+
},
|
| 2907 |
+
{
|
| 2908 |
+
"epoch": 10.236842105263158,
|
| 2909 |
+
"grad_norm": 14.182411193847656,
|
| 2910 |
+
"learning_rate": 8.170731707317074e-06,
|
| 2911 |
+
"loss": 0.0674,
|
| 2912 |
+
"step": 389
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 10.263157894736842,
|
| 2916 |
+
"grad_norm": 17.048315048217773,
|
| 2917 |
+
"learning_rate": 8.048780487804879e-06,
|
| 2918 |
+
"loss": 0.0707,
|
| 2919 |
+
"step": 390
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 10.289473684210526,
|
| 2923 |
+
"grad_norm": 6.421445369720459,
|
| 2924 |
+
"learning_rate": 7.926829268292683e-06,
|
| 2925 |
+
"loss": 0.0425,
|
| 2926 |
+
"step": 391
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"epoch": 10.31578947368421,
|
| 2930 |
+
"grad_norm": 6.346024036407471,
|
| 2931 |
+
"learning_rate": 7.804878048780489e-06,
|
| 2932 |
+
"loss": 0.0476,
|
| 2933 |
+
"step": 392
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 10.342105263157896,
|
| 2937 |
+
"grad_norm": 7.961727142333984,
|
| 2938 |
+
"learning_rate": 7.682926829268293e-06,
|
| 2939 |
+
"loss": 0.0716,
|
| 2940 |
+
"step": 393
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 10.368421052631579,
|
| 2944 |
+
"grad_norm": 5.910510063171387,
|
| 2945 |
+
"learning_rate": 7.560975609756098e-06,
|
| 2946 |
+
"loss": 0.0519,
|
| 2947 |
+
"step": 394
|
| 2948 |
+
},
|
| 2949 |
+
{
|
| 2950 |
+
"epoch": 10.394736842105264,
|
| 2951 |
+
"grad_norm": 8.626511573791504,
|
| 2952 |
+
"learning_rate": 7.439024390243902e-06,
|
| 2953 |
+
"loss": 0.0905,
|
| 2954 |
+
"step": 395
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 10.421052631578947,
|
| 2958 |
+
"grad_norm": 117.65007781982422,
|
| 2959 |
+
"learning_rate": 7.317073170731707e-06,
|
| 2960 |
+
"loss": 0.1747,
|
| 2961 |
+
"step": 396
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 10.447368421052632,
|
| 2965 |
+
"grad_norm": 50.54420471191406,
|
| 2966 |
+
"learning_rate": 7.195121951219512e-06,
|
| 2967 |
+
"loss": 0.4497,
|
| 2968 |
+
"step": 397
|
| 2969 |
+
},
|
| 2970 |
+
{
|
| 2971 |
+
"epoch": 10.473684210526315,
|
| 2972 |
+
"grad_norm": 106.56475830078125,
|
| 2973 |
+
"learning_rate": 7.073170731707317e-06,
|
| 2974 |
+
"loss": 0.2775,
|
| 2975 |
+
"step": 398
|
| 2976 |
+
},
|
| 2977 |
+
{
|
| 2978 |
+
"epoch": 10.5,
|
| 2979 |
+
"grad_norm": 68.34547424316406,
|
| 2980 |
+
"learning_rate": 6.951219512195123e-06,
|
| 2981 |
+
"loss": 0.1658,
|
| 2982 |
+
"step": 399
|
| 2983 |
+
},
|
| 2984 |
+
{
|
| 2985 |
+
"epoch": 10.526315789473685,
|
| 2986 |
+
"grad_norm": 37.34225082397461,
|
| 2987 |
+
"learning_rate": 6.829268292682928e-06,
|
| 2988 |
+
"loss": 0.1699,
|
| 2989 |
+
"step": 400
|
| 2990 |
+
},
|
| 2991 |
+
{
|
| 2992 |
+
"epoch": 10.552631578947368,
|
| 2993 |
+
"grad_norm": 35.83088302612305,
|
| 2994 |
+
"learning_rate": 6.707317073170733e-06,
|
| 2995 |
+
"loss": 0.1478,
|
| 2996 |
+
"step": 401
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 10.578947368421053,
|
| 3000 |
+
"grad_norm": 5.4022064208984375,
|
| 3001 |
+
"learning_rate": 6.585365853658537e-06,
|
| 3002 |
+
"loss": 0.0508,
|
| 3003 |
+
"step": 402
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 10.605263157894736,
|
| 3007 |
+
"grad_norm": 78.61613464355469,
|
| 3008 |
+
"learning_rate": 6.463414634146342e-06,
|
| 3009 |
+
"loss": 0.1148,
|
| 3010 |
+
"step": 403
|
| 3011 |
+
},
|
| 3012 |
+
{
|
| 3013 |
+
"epoch": 10.631578947368421,
|
| 3014 |
+
"grad_norm": 25.109567642211914,
|
| 3015 |
+
"learning_rate": 6.341463414634147e-06,
|
| 3016 |
+
"loss": 0.2027,
|
| 3017 |
+
"step": 404
|
| 3018 |
+
},
|
| 3019 |
+
{
|
| 3020 |
+
"epoch": 10.657894736842106,
|
| 3021 |
+
"grad_norm": 20.838315963745117,
|
| 3022 |
+
"learning_rate": 6.219512195121951e-06,
|
| 3023 |
+
"loss": 0.2343,
|
| 3024 |
+
"step": 405
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 10.68421052631579,
|
| 3028 |
+
"grad_norm": 18.96240997314453,
|
| 3029 |
+
"learning_rate": 6.0975609756097564e-06,
|
| 3030 |
+
"loss": 0.0477,
|
| 3031 |
+
"step": 406
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 10.710526315789474,
|
| 3035 |
+
"grad_norm": 5.693024158477783,
|
| 3036 |
+
"learning_rate": 5.975609756097561e-06,
|
| 3037 |
+
"loss": 0.0536,
|
| 3038 |
+
"step": 407
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 10.736842105263158,
|
| 3042 |
+
"grad_norm": 48.42964553833008,
|
| 3043 |
+
"learning_rate": 5.853658536585367e-06,
|
| 3044 |
+
"loss": 0.2161,
|
| 3045 |
+
"step": 408
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 10.763157894736842,
|
| 3049 |
+
"grad_norm": 14.99223518371582,
|
| 3050 |
+
"learning_rate": 5.731707317073172e-06,
|
| 3051 |
+
"loss": 0.0545,
|
| 3052 |
+
"step": 409
|
| 3053 |
+
},
|
| 3054 |
+
{
|
| 3055 |
+
"epoch": 10.789473684210526,
|
| 3056 |
+
"grad_norm": 22.92984390258789,
|
| 3057 |
+
"learning_rate": 5.609756097560976e-06,
|
| 3058 |
+
"loss": 0.0871,
|
| 3059 |
+
"step": 410
|
| 3060 |
+
},
|
| 3061 |
+
{
|
| 3062 |
+
"epoch": 10.81578947368421,
|
| 3063 |
+
"grad_norm": 35.50889205932617,
|
| 3064 |
+
"learning_rate": 5.487804878048781e-06,
|
| 3065 |
+
"loss": 0.2263,
|
| 3066 |
+
"step": 411
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 10.842105263157894,
|
| 3070 |
+
"grad_norm": 29.516984939575195,
|
| 3071 |
+
"learning_rate": 5.365853658536585e-06,
|
| 3072 |
+
"loss": 0.0998,
|
| 3073 |
+
"step": 412
|
| 3074 |
+
},
|
| 3075 |
+
{
|
| 3076 |
+
"epoch": 10.868421052631579,
|
| 3077 |
+
"grad_norm": 14.31261920928955,
|
| 3078 |
+
"learning_rate": 5.24390243902439e-06,
|
| 3079 |
+
"loss": 0.066,
|
| 3080 |
+
"step": 413
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 10.894736842105264,
|
| 3084 |
+
"grad_norm": 11.149723052978516,
|
| 3085 |
+
"learning_rate": 5.121951219512195e-06,
|
| 3086 |
+
"loss": 0.0432,
|
| 3087 |
+
"step": 414
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 10.921052631578947,
|
| 3091 |
+
"grad_norm": 7.057794570922852,
|
| 3092 |
+
"learning_rate": 5e-06,
|
| 3093 |
+
"loss": 0.0423,
|
| 3094 |
+
"step": 415
|
| 3095 |
+
},
|
| 3096 |
+
{
|
| 3097 |
+
"epoch": 10.947368421052632,
|
| 3098 |
+
"grad_norm": 15.57015609741211,
|
| 3099 |
+
"learning_rate": 4.8780487804878055e-06,
|
| 3100 |
+
"loss": 0.0554,
|
| 3101 |
+
"step": 416
|
| 3102 |
+
},
|
| 3103 |
+
{
|
| 3104 |
+
"epoch": 10.973684210526315,
|
| 3105 |
+
"grad_norm": 12.00108814239502,
|
| 3106 |
+
"learning_rate": 4.75609756097561e-06,
|
| 3107 |
+
"loss": 0.0587,
|
| 3108 |
+
"step": 417
|
| 3109 |
+
},
|
| 3110 |
+
{
|
| 3111 |
+
"epoch": 11.0,
|
| 3112 |
+
"grad_norm": 8.689730644226074,
|
| 3113 |
+
"learning_rate": 4.634146341463415e-06,
|
| 3114 |
+
"loss": 0.0436,
|
| 3115 |
+
"step": 418
|
| 3116 |
+
},
|
| 3117 |
+
{
|
| 3118 |
+
"epoch": 11.0,
|
| 3119 |
+
"eval_accuracy": 0.7,
|
| 3120 |
+
"eval_f1_macro": 0.6974548440065682,
|
| 3121 |
+
"eval_f1_micro": 0.7,
|
| 3122 |
+
"eval_f1_weighted": 0.6974548440065681,
|
| 3123 |
+
"eval_loss": 0.7847242951393127,
|
| 3124 |
+
"eval_precision_macro": 0.7133699633699633,
|
| 3125 |
+
"eval_precision_micro": 0.7,
|
| 3126 |
+
"eval_precision_weighted": 0.7133699633699634,
|
| 3127 |
+
"eval_recall_macro": 0.7,
|
| 3128 |
+
"eval_recall_micro": 0.7,
|
| 3129 |
+
"eval_recall_weighted": 0.7,
|
| 3130 |
+
"eval_runtime": 68.6184,
|
| 3131 |
+
"eval_samples_per_second": 0.874,
|
| 3132 |
+
"eval_steps_per_second": 0.029,
|
| 3133 |
+
"step": 418
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 11.026315789473685,
|
| 3137 |
+
"grad_norm": 15.764579772949219,
|
| 3138 |
+
"learning_rate": 4.51219512195122e-06,
|
| 3139 |
+
"loss": 0.0385,
|
| 3140 |
+
"step": 419
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 11.052631578947368,
|
| 3144 |
+
"grad_norm": 29.863691329956055,
|
| 3145 |
+
"learning_rate": 4.390243902439024e-06,
|
| 3146 |
+
"loss": 0.1211,
|
| 3147 |
+
"step": 420
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 11.078947368421053,
|
| 3151 |
+
"grad_norm": 3.3814544677734375,
|
| 3152 |
+
"learning_rate": 4.26829268292683e-06,
|
| 3153 |
+
"loss": 0.0295,
|
| 3154 |
+
"step": 421
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 11.105263157894736,
|
| 3158 |
+
"grad_norm": 57.14234924316406,
|
| 3159 |
+
"learning_rate": 4.146341463414634e-06,
|
| 3160 |
+
"loss": 0.0529,
|
| 3161 |
+
"step": 422
|
| 3162 |
+
},
|
| 3163 |
+
{
|
| 3164 |
+
"epoch": 11.131578947368421,
|
| 3165 |
+
"grad_norm": 11.074874877929688,
|
| 3166 |
+
"learning_rate": 4.024390243902439e-06,
|
| 3167 |
+
"loss": 0.0904,
|
| 3168 |
+
"step": 423
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 11.157894736842104,
|
| 3172 |
+
"grad_norm": 7.610099792480469,
|
| 3173 |
+
"learning_rate": 3.902439024390244e-06,
|
| 3174 |
+
"loss": 0.0426,
|
| 3175 |
+
"step": 424
|
| 3176 |
+
},
|
| 3177 |
+
{
|
| 3178 |
+
"epoch": 11.18421052631579,
|
| 3179 |
+
"grad_norm": 39.320194244384766,
|
| 3180 |
+
"learning_rate": 3.780487804878049e-06,
|
| 3181 |
+
"loss": 0.0485,
|
| 3182 |
+
"step": 425
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 11.210526315789474,
|
| 3186 |
+
"grad_norm": 7.378133773803711,
|
| 3187 |
+
"learning_rate": 3.6585365853658537e-06,
|
| 3188 |
+
"loss": 0.0408,
|
| 3189 |
+
"step": 426
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 11.236842105263158,
|
| 3193 |
+
"grad_norm": 27.179210662841797,
|
| 3194 |
+
"learning_rate": 3.5365853658536583e-06,
|
| 3195 |
+
"loss": 0.0803,
|
| 3196 |
+
"step": 427
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 11.263157894736842,
|
| 3200 |
+
"grad_norm": 9.848379135131836,
|
| 3201 |
+
"learning_rate": 3.414634146341464e-06,
|
| 3202 |
+
"loss": 0.066,
|
| 3203 |
+
"step": 428
|
| 3204 |
+
},
|
| 3205 |
+
{
|
| 3206 |
+
"epoch": 11.289473684210526,
|
| 3207 |
+
"grad_norm": 48.25758743286133,
|
| 3208 |
+
"learning_rate": 3.2926829268292685e-06,
|
| 3209 |
+
"loss": 0.1043,
|
| 3210 |
+
"step": 429
|
| 3211 |
+
},
|
| 3212 |
+
{
|
| 3213 |
+
"epoch": 11.31578947368421,
|
| 3214 |
+
"grad_norm": 87.25733184814453,
|
| 3215 |
+
"learning_rate": 3.1707317073170736e-06,
|
| 3216 |
+
"loss": 0.089,
|
| 3217 |
+
"step": 430
|
| 3218 |
+
},
|
| 3219 |
+
{
|
| 3220 |
+
"epoch": 11.342105263157896,
|
| 3221 |
+
"grad_norm": 5.009243965148926,
|
| 3222 |
+
"learning_rate": 3.0487804878048782e-06,
|
| 3223 |
+
"loss": 0.0378,
|
| 3224 |
+
"step": 431
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 11.368421052631579,
|
| 3228 |
+
"grad_norm": 27.693439483642578,
|
| 3229 |
+
"learning_rate": 2.9268292682926833e-06,
|
| 3230 |
+
"loss": 0.1046,
|
| 3231 |
+
"step": 432
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 11.394736842105264,
|
| 3235 |
+
"grad_norm": 20.024911880493164,
|
| 3236 |
+
"learning_rate": 2.804878048780488e-06,
|
| 3237 |
+
"loss": 0.0678,
|
| 3238 |
+
"step": 433
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 11.421052631578947,
|
| 3242 |
+
"grad_norm": 15.288347244262695,
|
| 3243 |
+
"learning_rate": 2.6829268292682926e-06,
|
| 3244 |
+
"loss": 0.0377,
|
| 3245 |
+
"step": 434
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 11.447368421052632,
|
| 3249 |
+
"grad_norm": 17.69446563720703,
|
| 3250 |
+
"learning_rate": 2.5609756097560977e-06,
|
| 3251 |
+
"loss": 0.0706,
|
| 3252 |
+
"step": 435
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 11.473684210526315,
|
| 3256 |
+
"grad_norm": 21.541053771972656,
|
| 3257 |
+
"learning_rate": 2.4390243902439027e-06,
|
| 3258 |
+
"loss": 0.128,
|
| 3259 |
+
"step": 436
|
| 3260 |
+
},
|
| 3261 |
+
{
|
| 3262 |
+
"epoch": 11.5,
|
| 3263 |
+
"grad_norm": 15.488064765930176,
|
| 3264 |
+
"learning_rate": 2.3170731707317074e-06,
|
| 3265 |
+
"loss": 0.0361,
|
| 3266 |
+
"step": 437
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 11.526315789473685,
|
| 3270 |
+
"grad_norm": 22.015546798706055,
|
| 3271 |
+
"learning_rate": 2.195121951219512e-06,
|
| 3272 |
+
"loss": 0.1368,
|
| 3273 |
+
"step": 438
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 11.552631578947368,
|
| 3277 |
+
"grad_norm": 11.265583038330078,
|
| 3278 |
+
"learning_rate": 2.073170731707317e-06,
|
| 3279 |
+
"loss": 0.0636,
|
| 3280 |
+
"step": 439
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 11.578947368421053,
|
| 3284 |
+
"grad_norm": 21.78879165649414,
|
| 3285 |
+
"learning_rate": 1.951219512195122e-06,
|
| 3286 |
+
"loss": 0.3094,
|
| 3287 |
+
"step": 440
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 11.605263157894736,
|
| 3291 |
+
"grad_norm": 25.989824295043945,
|
| 3292 |
+
"learning_rate": 1.8292682926829268e-06,
|
| 3293 |
+
"loss": 0.0661,
|
| 3294 |
+
"step": 441
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 11.631578947368421,
|
| 3298 |
+
"grad_norm": 18.91653060913086,
|
| 3299 |
+
"learning_rate": 1.707317073170732e-06,
|
| 3300 |
+
"loss": 0.0713,
|
| 3301 |
+
"step": 442
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 11.657894736842106,
|
| 3305 |
+
"grad_norm": 5.307289123535156,
|
| 3306 |
+
"learning_rate": 1.5853658536585368e-06,
|
| 3307 |
+
"loss": 0.04,
|
| 3308 |
+
"step": 443
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 11.68421052631579,
|
| 3312 |
+
"grad_norm": 19.790300369262695,
|
| 3313 |
+
"learning_rate": 1.4634146341463416e-06,
|
| 3314 |
+
"loss": 0.0798,
|
| 3315 |
+
"step": 444
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 11.710526315789474,
|
| 3319 |
+
"grad_norm": 30.500885009765625,
|
| 3320 |
+
"learning_rate": 1.3414634146341463e-06,
|
| 3321 |
+
"loss": 0.1807,
|
| 3322 |
+
"step": 445
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 11.736842105263158,
|
| 3326 |
+
"grad_norm": 18.471010208129883,
|
| 3327 |
+
"learning_rate": 1.2195121951219514e-06,
|
| 3328 |
+
"loss": 0.1433,
|
| 3329 |
+
"step": 446
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 11.763157894736842,
|
| 3333 |
+
"grad_norm": 4.643721103668213,
|
| 3334 |
+
"learning_rate": 1.097560975609756e-06,
|
| 3335 |
+
"loss": 0.0299,
|
| 3336 |
+
"step": 447
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 11.789473684210526,
|
| 3340 |
+
"grad_norm": 15.291370391845703,
|
| 3341 |
+
"learning_rate": 9.75609756097561e-07,
|
| 3342 |
+
"loss": 0.0583,
|
| 3343 |
+
"step": 448
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 11.81578947368421,
|
| 3347 |
+
"grad_norm": 21.351959228515625,
|
| 3348 |
+
"learning_rate": 8.53658536585366e-07,
|
| 3349 |
+
"loss": 0.0581,
|
| 3350 |
+
"step": 449
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 11.842105263157894,
|
| 3354 |
+
"grad_norm": 39.182071685791016,
|
| 3355 |
+
"learning_rate": 7.317073170731708e-07,
|
| 3356 |
+
"loss": 0.1439,
|
| 3357 |
+
"step": 450
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 11.868421052631579,
|
| 3361 |
+
"grad_norm": 3.376581907272339,
|
| 3362 |
+
"learning_rate": 6.097560975609757e-07,
|
| 3363 |
+
"loss": 0.0393,
|
| 3364 |
+
"step": 451
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 11.894736842105264,
|
| 3368 |
+
"grad_norm": 20.73614501953125,
|
| 3369 |
+
"learning_rate": 4.878048780487805e-07,
|
| 3370 |
+
"loss": 0.0681,
|
| 3371 |
+
"step": 452
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 11.921052631578947,
|
| 3375 |
+
"grad_norm": 11.527457237243652,
|
| 3376 |
+
"learning_rate": 3.658536585365854e-07,
|
| 3377 |
+
"loss": 0.0408,
|
| 3378 |
+
"step": 453
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 11.947368421052632,
|
| 3382 |
+
"grad_norm": 12.897309303283691,
|
| 3383 |
+
"learning_rate": 2.439024390243903e-07,
|
| 3384 |
+
"loss": 0.0337,
|
| 3385 |
+
"step": 454
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 11.973684210526315,
|
| 3389 |
+
"grad_norm": 39.30110549926758,
|
| 3390 |
+
"learning_rate": 1.2195121951219514e-07,
|
| 3391 |
+
"loss": 0.1508,
|
| 3392 |
+
"step": 455
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 12.0,
|
| 3396 |
+
"grad_norm": 106.00151062011719,
|
| 3397 |
+
"learning_rate": 0.0,
|
| 3398 |
+
"loss": 0.042,
|
| 3399 |
+
"step": 456
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 12.0,
|
| 3403 |
+
"eval_accuracy": 0.7666666666666667,
|
| 3404 |
+
"eval_f1_macro": 0.7635683760683761,
|
| 3405 |
+
"eval_f1_micro": 0.7666666666666667,
|
| 3406 |
+
"eval_f1_weighted": 0.763568376068376,
|
| 3407 |
+
"eval_loss": 0.7394976019859314,
|
| 3408 |
+
"eval_precision_macro": 0.7832887700534759,
|
| 3409 |
+
"eval_precision_micro": 0.7666666666666667,
|
| 3410 |
+
"eval_precision_weighted": 0.783288770053476,
|
| 3411 |
+
"eval_recall_macro": 0.7666666666666666,
|
| 3412 |
+
"eval_recall_micro": 0.7666666666666667,
|
| 3413 |
+
"eval_recall_weighted": 0.7666666666666667,
|
| 3414 |
+
"eval_runtime": 65.796,
|
| 3415 |
+
"eval_samples_per_second": 0.912,
|
| 3416 |
+
"eval_steps_per_second": 0.03,
|
| 3417 |
+
"step": 456
|
| 3418 |
+
}
|
| 3419 |
+
],
|
| 3420 |
+
"logging_steps": 1,
|
| 3421 |
+
"max_steps": 456,
|
| 3422 |
+
"num_input_tokens_seen": 0,
|
| 3423 |
+
"num_train_epochs": 12,
|
| 3424 |
+
"save_steps": 500,
|
| 3425 |
+
"stateful_callbacks": {
|
| 3426 |
+
"EarlyStoppingCallback": {
|
| 3427 |
+
"args": {
|
| 3428 |
+
"early_stopping_patience": 5,
|
| 3429 |
+
"early_stopping_threshold": 0.01
|
| 3430 |
+
},
|
| 3431 |
+
"attributes": {
|
| 3432 |
+
"early_stopping_patience_counter": 0
|
| 3433 |
+
}
|
| 3434 |
+
},
|
| 3435 |
+
"TrainerControl": {
|
| 3436 |
+
"args": {
|
| 3437 |
+
"should_epoch_stop": false,
|
| 3438 |
+
"should_evaluate": false,
|
| 3439 |
+
"should_log": false,
|
| 3440 |
+
"should_save": true,
|
| 3441 |
+
"should_training_stop": true
|
| 3442 |
+
},
|
| 3443 |
+
"attributes": {}
|
| 3444 |
+
}
|
| 3445 |
+
},
|
| 3446 |
+
"total_flos": 171813516853248.0,
|
| 3447 |
+
"train_batch_size": 16,
|
| 3448 |
+
"trial_name": null,
|
| 3449 |
+
"trial_params": null
|
| 3450 |
+
}
|
checkpoint-456/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:488d952d2b2f4757986d04ba3081d970de0534335a7e9cc2f9e5cba483eb2afb
|
| 3 |
+
size 5368
|
config.json
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "albert/albert-base-v2",
|
| 3 |
+
"_num_labels": 4,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"AlbertForSequenceClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_probs_dropout_prob": 0,
|
| 8 |
+
"bos_token_id": 2,
|
| 9 |
+
"classifier_dropout_prob": 0.1,
|
| 10 |
+
"down_scale_factor": 1,
|
| 11 |
+
"embedding_size": 128,
|
| 12 |
+
"eos_token_id": 3,
|
| 13 |
+
"gap_size": 0,
|
| 14 |
+
"hidden_act": "gelu_new",
|
| 15 |
+
"hidden_dropout_prob": 0,
|
| 16 |
+
"hidden_size": 768,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "CLE_DSI_SIEP",
|
| 19 |
+
"1": "CLE_DSI_SIL",
|
| 20 |
+
"2": "CLE_DSI_SPIA",
|
| 21 |
+
"3": "CLE_DSI_SSUR"
|
| 22 |
+
},
|
| 23 |
+
"initializer_range": 0.02,
|
| 24 |
+
"inner_group_num": 1,
|
| 25 |
+
"intermediate_size": 3072,
|
| 26 |
+
"label2id": {
|
| 27 |
+
"CLE_DSI_SIEP": 0,
|
| 28 |
+
"CLE_DSI_SIL": 1,
|
| 29 |
+
"CLE_DSI_SPIA": 2,
|
| 30 |
+
"CLE_DSI_SSUR": 3
|
| 31 |
+
},
|
| 32 |
+
"layer_norm_eps": 1e-12,
|
| 33 |
+
"max_position_embeddings": 512,
|
| 34 |
+
"model_type": "albert",
|
| 35 |
+
"net_structure_type": 0,
|
| 36 |
+
"num_attention_heads": 12,
|
| 37 |
+
"num_hidden_groups": 1,
|
| 38 |
+
"num_hidden_layers": 12,
|
| 39 |
+
"num_memory_blocks": 0,
|
| 40 |
+
"pad_token_id": 0,
|
| 41 |
+
"position_embedding_type": "absolute",
|
| 42 |
+
"problem_type": "single_label_classification",
|
| 43 |
+
"torch_dtype": "float32",
|
| 44 |
+
"transformers_version": "4.48.0",
|
| 45 |
+
"type_vocab_size": 2,
|
| 46 |
+
"vocab_size": 30000
|
| 47 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7de090cbcc2fc88c209787820c086224988f028790b65f74b65f19186c33681f
|
| 3 |
+
size 46750064
|
runs/Jun30_09-14-46_r-bastien2-gt-inno-2d039o9i-771b3-8acwj/events.out.tfevents.1751274888.r-bastien2-gt-inno-2d039o9i-771b3-8acwj.108.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f222dc62274514b7c2935f5ef5c155224734529f4bf66f5aa5c6d8c266b474c8
|
| 3 |
+
size 111511
|
runs/Jun30_09-14-46_r-bastien2-gt-inno-2d039o9i-771b3-8acwj/events.out.tfevents.1751302564.r-bastien2-gt-inno-2d039o9i-771b3-8acwj.108.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a52c7e1ab09c533e8cf0023a3baa11d093066a3a4ffa209e4729e8e35366048
|
| 3 |
+
size 921
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"eos_token": "[SEP]",
|
| 5 |
+
"mask_token": {
|
| 6 |
+
"content": "[MASK]",
|
| 7 |
+
"lstrip": true,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"pad_token": "<pad>",
|
| 13 |
+
"sep_token": "[SEP]",
|
| 14 |
+
"unk_token": "<unk>"
|
| 15 |
+
}
|
spiece.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fefb02b667a6c5c2fe27602d28e5fb3428f66ab89c7d6f388e7c8d44a02d0336
|
| 3 |
+
size 760289
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<pad>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<unk>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"4": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": true,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "[CLS]",
|
| 45 |
+
"clean_up_tokenization_spaces": false,
|
| 46 |
+
"cls_token": "[CLS]",
|
| 47 |
+
"do_lower_case": true,
|
| 48 |
+
"eos_token": "[SEP]",
|
| 49 |
+
"extra_special_tokens": {},
|
| 50 |
+
"keep_accents": false,
|
| 51 |
+
"mask_token": "[MASK]",
|
| 52 |
+
"model_max_length": 512,
|
| 53 |
+
"pad_token": "<pad>",
|
| 54 |
+
"remove_space": true,
|
| 55 |
+
"sep_token": "[SEP]",
|
| 56 |
+
"tokenizer_class": "AlbertTokenizer",
|
| 57 |
+
"unk_token": "<unk>"
|
| 58 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:488d952d2b2f4757986d04ba3081d970de0534335a7e9cc2f9e5cba483eb2afb
|
| 3 |
+
size 5368
|
training_params.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "MerlAIn-Base-Albert-005/autotrain-data",
|
| 3 |
+
"model": "albert/albert-base-v2",
|
| 4 |
+
"lr": 5e-05,
|
| 5 |
+
"epochs": 12,
|
| 6 |
+
"max_seq_length": 512,
|
| 7 |
+
"batch_size": 16,
|
| 8 |
+
"warmup_ratio": 0.1,
|
| 9 |
+
"gradient_accumulation": 1,
|
| 10 |
+
"optimizer": "adamw_torch",
|
| 11 |
+
"scheduler": "linear",
|
| 12 |
+
"weight_decay": 0.0,
|
| 13 |
+
"max_grad_norm": 1.0,
|
| 14 |
+
"seed": 42,
|
| 15 |
+
"train_split": "train",
|
| 16 |
+
"valid_split": "validation",
|
| 17 |
+
"text_column": "autotrain_text",
|
| 18 |
+
"target_column": "autotrain_label",
|
| 19 |
+
"logging_steps": -1,
|
| 20 |
+
"project_name": "MerlAIn-Base-Albert-005",
|
| 21 |
+
"auto_find_batch_size": false,
|
| 22 |
+
"mixed_precision": "fp16",
|
| 23 |
+
"save_total_limit": 1,
|
| 24 |
+
"push_to_hub": true,
|
| 25 |
+
"eval_strategy": "epoch",
|
| 26 |
+
"username": "Bastien2",
|
| 27 |
+
"log": "tensorboard",
|
| 28 |
+
"early_stopping_patience": 5,
|
| 29 |
+
"early_stopping_threshold": 0.01
|
| 30 |
+
}
|