Model save

README.md

@@ -0,0 +1,61 @@
+---
+library_name: transformers
+license: apache-2.0
+base_model: alignment-handbook/zephyr-7b-sft-full
+tags:
+- trl
+- dpo
+- generated_from_trainer
+model-index:
+- name: zephyr-7b-align-scan-0.0-0.8-cosine-1
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# zephyr-7b-align-scan-0.0-0.8-cosine-1
+
+This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 3.662370190418977e-07
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 4
+- gradient_accumulation_steps: 2
+- total_train_batch_size: 64
+- total_eval_batch_size: 32
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 1
+
+### Training results
+
+
+
+### Framework versions
+
+- Transformers 4.44.2
+- Pytorch 2.4.0
+- Datasets 2.21.0
+- Tokenizers 0.19.1

all_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 1.0,
+    "total_flos": 0.0,
+    "train_loss": 0.6724727004766464,
+    "train_runtime": 961.1077,
+    "train_samples": 6113,
+    "train_samples_per_second": 6.36,
+    "train_steps_per_second": 0.1
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "transformers_version": "4.44.2"
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 1.0,
+    "total_flos": 0.0,
+    "train_loss": 0.6724727004766464,
+    "train_runtime": 961.1077,
+    "train_samples": 6113,
+    "train_samples_per_second": 6.36,
+    "train_steps_per_second": 0.1
+}

trainer_state.json

@@ -0,0 +1,192 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 100,
+  "global_step": 96,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.010416666666666666,
+      "grad_norm": 534.9856766765029,
+      "learning_rate": 3.6623701904189776e-08,
+      "logits/chosen": -2.590585231781006,
+      "logits/rejected": -2.5664222240448,
+      "logps/chosen": -80.29847717285156,
+      "logps/rejected": -53.10200881958008,
+      "loss": 0.6931,
+      "rewards/accuracies": 0.0,
+      "rewards/chosen": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/rejected": 0.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.10416666666666667,
+      "grad_norm": 522.7071434903489,
+      "learning_rate": 3.662370190418977e-07,
+      "logits/chosen": -2.5560922622680664,
+      "logits/rejected": -2.5382773876190186,
+      "logps/chosen": -87.9105224609375,
+      "logps/rejected": -81.0152587890625,
+      "loss": 0.7035,
+      "rewards/accuracies": 0.2152777761220932,
+      "rewards/chosen": 0.0037625303957611322,
+      "rewards/margins": -0.015120850875973701,
+      "rewards/rejected": 0.01888338290154934,
+      "step": 10
+    },
+    {
+      "epoch": 0.20833333333333334,
+      "grad_norm": 330.8070947624322,
+      "learning_rate": 3.5415414722060956e-07,
+      "logits/chosen": -2.6099281311035156,
+      "logits/rejected": -2.5609025955200195,
+      "logps/chosen": -102.80690002441406,
+      "logps/rejected": -89.45201110839844,
+      "loss": 0.6595,
+      "rewards/accuracies": 0.36250001192092896,
+      "rewards/chosen": 0.4419492781162262,
+      "rewards/margins": 0.09075690805912018,
+      "rewards/rejected": 0.3511923849582672,
+      "step": 20
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 561.0026948523158,
+      "learning_rate": 3.1950008155765393e-07,
+      "logits/chosen": -2.510303020477295,
+      "logits/rejected": -2.5249061584472656,
+      "logps/chosen": -67.08929443359375,
+      "logps/rejected": -75.05818176269531,
+      "loss": 0.6859,
+      "rewards/accuracies": 0.3375000059604645,
+      "rewards/chosen": 0.2826586067676544,
+      "rewards/margins": 0.39012494683265686,
+      "rewards/rejected": -0.10746632516384125,
+      "step": 30
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 402.3919923341537,
+      "learning_rate": 2.6684804225439007e-07,
+      "logits/chosen": -2.5768940448760986,
+      "logits/rejected": -2.5649542808532715,
+      "logps/chosen": -72.13874053955078,
+      "logps/rejected": -70.94956970214844,
+      "loss": 0.6997,
+      "rewards/accuracies": 0.25,
+      "rewards/chosen": 0.00632839510217309,
+      "rewards/margins": 0.19826939702033997,
+      "rewards/rejected": -0.19194099307060242,
+      "step": 40
+    },
+    {
+      "epoch": 0.5208333333333334,
+      "grad_norm": 382.89112989307444,
+      "learning_rate": 2.0314640218361037e-07,
+      "logits/chosen": -2.494668960571289,
+      "logits/rejected": -2.5067570209503174,
+      "logps/chosen": -50.37248229980469,
+      "logps/rejected": -58.68071746826172,
+      "loss": 0.695,
+      "rewards/accuracies": 0.24375000596046448,
+      "rewards/chosen": 0.4294072091579437,
+      "rewards/margins": 0.24388039112091064,
+      "rewards/rejected": 0.1855268031358719,
+      "step": 50
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 356.73514211856235,
+      "learning_rate": 1.3680172552929206e-07,
+      "logits/chosen": -2.58482027053833,
+      "logits/rejected": -2.566230535507202,
+      "logps/chosen": -78.02713775634766,
+      "logps/rejected": -78.90872955322266,
+      "loss": 0.6665,
+      "rewards/accuracies": 0.3499999940395355,
+      "rewards/chosen": 0.8464384078979492,
+      "rewards/margins": 0.4981551766395569,
+      "rewards/rejected": 0.3482832610607147,
+      "step": 60
+    },
+    {
+      "epoch": 0.7291666666666666,
+      "grad_norm": 347.90887844338835,
+      "learning_rate": 7.656937215250093e-08,
+      "logits/chosen": -2.5671563148498535,
+      "logits/rejected": -2.558472156524658,
+      "logps/chosen": -95.0243148803711,
+      "logps/rejected": -84.78661346435547,
+      "loss": 0.672,
+      "rewards/accuracies": 0.34375,
+      "rewards/chosen": 1.0726714134216309,
+      "rewards/margins": 0.5336906313896179,
+      "rewards/rejected": 0.5389808416366577,
+      "step": 70
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 625.7311819069191,
+      "learning_rate": 3.039807216523069e-08,
+      "logits/chosen": -2.569451332092285,
+      "logits/rejected": -2.516343593597412,
+      "logps/chosen": -84.30049133300781,
+      "logps/rejected": -78.89418029785156,
+      "loss": 0.6324,
+      "rewards/accuracies": 0.41874998807907104,
+      "rewards/chosen": 1.0922737121582031,
+      "rewards/margins": 0.6956304311752319,
+      "rewards/rejected": 0.3966432213783264,
+      "step": 80
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 530.9459931532258,
+      "learning_rate": 4.380949643555316e-09,
+      "logits/chosen": -2.502892017364502,
+      "logits/rejected": -2.4977681636810303,
+      "logps/chosen": -53.26112747192383,
+      "logps/rejected": -63.1404914855957,
+      "loss": 0.6758,
+      "rewards/accuracies": 0.2562499940395355,
+      "rewards/chosen": 0.58380526304245,
+      "rewards/margins": 0.31763672828674316,
+      "rewards/rejected": 0.2661685645580292,
+      "step": 90
+    },
+    {
+      "epoch": 1.0,
+      "step": 96,
+      "total_flos": 0.0,
+      "train_loss": 0.6724727004766464,
+      "train_runtime": 961.1077,
+      "train_samples_per_second": 6.36,
+      "train_steps_per_second": 0.1
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 96,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}