Model save

README.md

@@ -2,15 +2,11 @@
 license: apache-2.0
 base_model: BioMistral/BioMistral-7B
 tags:
-- alignment-handbook
-- trl
-- sft
-- generated_from_trainer
 - trl
 - sft
 - generated_from_trainer
 datasets:
-- HuggingFaceH4/deita-10k-v0-sft
+- generator
 model-index:
 - name: biomistral-7b-wo-kqa_golden-iter-sft-step1
   results: []
@@ -21,9 +17,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # biomistral-7b-wo-kqa_golden-iter-sft-step1
 
-This model is a fine-tuned version of [BioMistral/BioMistral-7B](https://huggingface.co/BioMistral/BioMistral-7B) on the HuggingFaceH4/deita-10k-v0-sft dataset.
+This model is a fine-tuned version of [BioMistral/BioMistral-7B](https://huggingface.co/BioMistral/BioMistral-7B) on the generator dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.4295
+- Loss: 1.8395
 
 ## Model description
 
@@ -60,14 +56,14 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.8334        | 0.92  | 9    | 1.4131          |
-| 0.6303        | 1.95  | 19   | 1.4060          |
-| 0.4647        | 2.77  | 27   | 1.4295          |
+| 1.7589        | 1.0   | 7    | 1.7180          |
+| 1.4041        | 2.0   | 14   | 1.7635          |
+| 0.6946        | 3.0   | 21   | 1.8395          |
 
 
 ### Framework versions
 
-- Transformers 4.39.0.dev0
-- Pytorch 2.1.2
+- Transformers 4.38.2
+- Pytorch 2.1.2+cu121
 - Datasets 2.14.6
 - Tokenizers 0.15.2

all_results.json

@@ -1,13 +1,8 @@
 {
-    "epoch": 2.77,
-    "eval_loss": 1.4294962882995605,
-    "eval_runtime": 38.9103,
-    "eval_samples": 4044,
-    "eval_samples_per_second": 8.841,
-    "eval_steps_per_second": 0.565,
-    "train_loss": 0.6124309632513258,
-    "train_runtime": 676.0028,
+    "epoch": 3.0,
+    "train_loss": 1.1876622920944577,
+    "train_runtime": 427.6941,
     "train_samples": 4750,
-    "train_samples_per_second": 2.72,
-    "train_steps_per_second": 0.04
+    "train_samples_per_second": 3.065,
+    "train_steps_per_second": 0.049
 }

config.json

@@ -20,7 +20,7 @@
   "sliding_window": 4096,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.39.0.dev0",
-  "use_cache": true,
+  "transformers_version": "4.38.2",
+  "use_cache": false,
   "vocab_size": 32000
 }

generation_config.json

@@ -2,5 +2,5 @@
   "_from_model_config": true,
   "bos_token_id": 1,
   "eos_token_id": 2,
-  "transformers_version": "4.39.0.dev0"
+  "transformers_version": "4.38.2"
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:da75fcbb729830fc5bffe2a3dde01be33068b3c5d1138b6f9e6f17fd94b3dff8
+oid sha256:de60edc1d447a43d526f0a7cd8b5394d4fb25f2e8b54c99a2aaa483fbcea90f6
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:71e80be36b549dbfc42cbfe4fc54e3b51d0f03a4dc48a7f47646010492d2f314
+oid sha256:3152f1d3f5d1010b71b73360fed6f92fff5c249a609e464b53e61bb8ceeb2250
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:867985b25124dc090645cae581460d25989b30e7b2f668fc2769f6e68ded1b79
+oid sha256:6ff759b5b480a27ee7333f8a507cf8cf48f8d382938602357de1c39a583c5847
 size 4540516344

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.77,
-    "train_loss": 0.6124309632513258,
-    "train_runtime": 676.0028,
+    "epoch": 3.0,
+    "train_loss": 1.1876622920944577,
+    "train_runtime": 427.6941,
     "train_samples": 4750,
-    "train_samples_per_second": 2.72,
-    "train_steps_per_second": 0.04
+    "train_samples_per_second": 3.065,
+    "train_steps_per_second": 0.049
 }

trainer_state.json

@@ -1,95 +1,88 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.769230769230769,
+  "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 27,
+  "global_step": 21,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.1,
-      "grad_norm": 24.191700961175112,
+      "epoch": 0.14,
+      "grad_norm": 10.547944577355347,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.5581,
+      "loss": 1.7384,
       "step": 1
     },
     {
-      "epoch": 0.51,
-      "grad_norm": 12.576350745768869,
-      "learning_rate": 1.9659258262890683e-05,
-      "loss": 0.8334,
+      "epoch": 0.71,
+      "grad_norm": 13.625247977537818,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 1.7589,
       "step": 5
     },
     {
-      "epoch": 0.92,
-      "eval_loss": 1.4131343364715576,
-      "eval_runtime": 38.4185,
-      "eval_samples_per_second": 8.954,
-      "eval_steps_per_second": 0.573,
-      "step": 9
+      "epoch": 1.0,
+      "eval_loss": 1.7180463075637817,
+      "eval_runtime": 1.7598,
+      "eval_samples_per_second": 7.387,
+      "eval_steps_per_second": 0.568,
+      "step": 7
     },
     {
-      "epoch": 1.03,
-      "grad_norm": 17.837211332799143,
-      "learning_rate": 1.608761429008721e-05,
-      "loss": 0.738,
+      "epoch": 1.43,
+      "grad_norm": 5.744275140398176,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 1.4041,
       "step": 10
     },
     {
-      "epoch": 1.54,
-      "grad_norm": 3.797877129136344,
-      "learning_rate": 1e-05,
-      "loss": 0.6303,
-      "step": 15
+      "epoch": 2.0,
+      "eval_loss": 1.763495683670044,
+      "eval_runtime": 1.6856,
+      "eval_samples_per_second": 7.712,
+      "eval_steps_per_second": 0.593,
+      "step": 14
     },
     {
-      "epoch": 1.95,
-      "eval_loss": 1.4060215950012207,
-      "eval_runtime": 39.241,
-      "eval_samples_per_second": 8.766,
-      "eval_steps_per_second": 0.561,
-      "step": 19
+      "epoch": 2.14,
+      "grad_norm": 7.096816156714056,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.0218,
+      "step": 15
     },
     {
-      "epoch": 2.05,
-      "grad_norm": 7.485764754081964,
-      "learning_rate": 3.912385709912794e-06,
-      "loss": 0.5348,
+      "epoch": 2.86,
+      "grad_norm": 4.77063469960506,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.6946,
       "step": 20
     },
     {
-      "epoch": 2.56,
-      "grad_norm": 3.0173543034517323,
-      "learning_rate": 3.4074173710931804e-07,
-      "loss": 0.4647,
-      "step": 25
-    },
-    {
-      "epoch": 2.77,
-      "eval_loss": 1.4294962882995605,
-      "eval_runtime": 38.6474,
-      "eval_samples_per_second": 8.901,
-      "eval_steps_per_second": 0.569,
-      "step": 27
+      "epoch": 3.0,
+      "eval_loss": 1.8394839763641357,
+      "eval_runtime": 1.7001,
+      "eval_samples_per_second": 7.647,
+      "eval_steps_per_second": 0.588,
+      "step": 21
     },
     {
-      "epoch": 2.77,
-      "step": 27,
-      "total_flos": 5600905789440.0,
-      "train_loss": 0.6124309632513258,
-      "train_runtime": 676.0028,
-      "train_samples_per_second": 2.72,
-      "train_steps_per_second": 0.04
+      "epoch": 3.0,
+      "step": 21,
+      "total_flos": 4344627855360.0,
+      "train_loss": 1.1876622920944577,
+      "train_runtime": 427.6941,
+      "train_samples_per_second": 3.065,
+      "train_steps_per_second": 0.049
     }
   ],
   "logging_steps": 5,
-  "max_steps": 27,
+  "max_steps": 21,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 500,
-  "total_flos": 5600905789440.0,
+  "total_flos": 4344627855360.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9605dab41a6f0bcc6542eda4f5de7efa4b0c431b8dd3c26e5d4c64263bed48ba
+oid sha256:6b19306777988d00a289fd79b99cc19d76222c0c39bf75362ee3f92ab44f2161
 size 6200