Training in progress, epoch 1
Browse files- .gitignore +1 -0
- config.json +32 -0
- logs/1681639877.9057024/events.out.tfevents.1681639877.user-SYS-4029GP-TRT2-EC028B.45297.1 +3 -0
- logs/1681640563.3552833/events.out.tfevents.1681640563.user-SYS-4029GP-TRT2-EC028B.69621.1 +3 -0
- logs/1681642917.3956923/events.out.tfevents.1681642917.user-SYS-4029GP-TRT2-EC028B.49005.1 +3 -0
- logs/events.out.tfevents.1681639877.user-SYS-4029GP-TRT2-EC028B.45297.0 +3 -0
- logs/events.out.tfevents.1681640563.user-SYS-4029GP-TRT2-EC028B.69621.0 +3 -0
- logs/events.out.tfevents.1681642917.user-SYS-4029GP-TRT2-EC028B.49005.0 +3 -0
- pytorch_model.bin +3 -0
- training_args.bin +3 -0
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
checkpoint-*/
|
config.json
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "google/flan-t5-large",
|
3 |
+
"architectures": [
|
4 |
+
"T5ForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"d_ff": 2816,
|
7 |
+
"d_kv": 64,
|
8 |
+
"d_model": 1024,
|
9 |
+
"decoder_start_token_id": 0,
|
10 |
+
"dense_act_fn": "gelu_new",
|
11 |
+
"dropout_rate": 0.1,
|
12 |
+
"eos_token_id": 1,
|
13 |
+
"feed_forward_proj": "gated-gelu",
|
14 |
+
"initializer_factor": 1.0,
|
15 |
+
"is_encoder_decoder": true,
|
16 |
+
"is_gated_act": true,
|
17 |
+
"layer_norm_epsilon": 1e-06,
|
18 |
+
"model_type": "t5",
|
19 |
+
"n_positions": 512,
|
20 |
+
"num_decoder_layers": 24,
|
21 |
+
"num_heads": 16,
|
22 |
+
"num_layers": 24,
|
23 |
+
"output_past": true,
|
24 |
+
"pad_token_id": 0,
|
25 |
+
"relative_attention_max_distance": 128,
|
26 |
+
"relative_attention_num_buckets": 32,
|
27 |
+
"tie_word_embeddings": false,
|
28 |
+
"torch_dtype": "float32",
|
29 |
+
"transformers_version": "4.26.0",
|
30 |
+
"use_cache": false,
|
31 |
+
"vocab_size": 32128
|
32 |
+
}
|
logs/1681639877.9057024/events.out.tfevents.1681639877.user-SYS-4029GP-TRT2-EC028B.45297.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9f2cb124a885ee6450bf36226f410750e01a6d7ba1a4b9df94f6acc8749c566d
|
3 |
+
size 5957
|
logs/1681640563.3552833/events.out.tfevents.1681640563.user-SYS-4029GP-TRT2-EC028B.69621.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:eb499737ca71d666b649d1505fab716562522a3510ce1c97b427f5c3a8d15a4e
|
3 |
+
size 5957
|
logs/1681642917.3956923/events.out.tfevents.1681642917.user-SYS-4029GP-TRT2-EC028B.49005.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0cfaf8464559cadb9093ac3ccce298196ae2c4ac89167a070d55546c42f5ba0
|
3 |
+
size 5957
|
logs/events.out.tfevents.1681639877.user-SYS-4029GP-TRT2-EC028B.45297.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a9f66331887a68a06fffc5583487512a700e840772148d6a9e4c69a4bf0f8dc0
|
3 |
+
size 4096
|
logs/events.out.tfevents.1681640563.user-SYS-4029GP-TRT2-EC028B.69621.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d11a151ae342e6b8b542b36eed271cc788ea8d5855a536fc0d595e38023d87bb
|
3 |
+
size 4253
|
logs/events.out.tfevents.1681642917.user-SYS-4029GP-TRT2-EC028B.49005.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:97ac2b7890c3655af1f28ae58bc008f2f514f5ddecff04895df54af6d3748cfa
|
3 |
+
size 5720
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:93909265a9a5d1bbd10a2c6dacc5f988c32a61bdb39fa5fd678bb25583cd00b4
|
3 |
+
size 3132789733
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1a291723a2bf8b0de9aae9f61ab8a4c165b2156b1ecee8d1263a50f843fa0edf
|
3 |
+
size 3695
|