Model save

README.md

@@ -3,7 +3,6 @@ base_model: meta-llama/Llama-3.1-8B-Instruct
 library_name: peft
 license: llama3.1
 tags:
-- alignment-handbook
 - trl
 - sft
 - generated_from_trainer
@@ -19,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [meta-llama/Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.5322
+- Loss: 1.5113
 
 ## Model description
 
@@ -50,16 +49,18 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- training_steps: 20
+- training_steps: 30
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 2.3938        | 0.7692 | 5    | 1.7584          |
-| 1.4879        | 1.5385 | 10   | 1.5969          |
-| 1.3838        | 2.3077 | 15   | 1.5445          |
-| 1.2126        | 3.0769 | 20   | 1.5322          |
+| 2.4578        | 0.7692 | 5    | 1.7601          |
+| 1.4794        | 1.5385 | 10   | 1.5840          |
+| 1.3439        | 2.3077 | 15   | 1.4991          |
+| 1.0898        | 3.0769 | 20   | 1.4789          |
+| 1.0223        | 3.8462 | 25   | 1.5093          |
+| 0.9906        | 4.6154 | 30   | 1.5113          |
 
 
 ### Framework versions

all_results.json

@@ -1,14 +1,9 @@
 {
-    "epoch": 3.076923076923077,
-    "eval_loss": 1.5321630239486694,
-    "eval_runtime": 1.2839,
-    "eval_samples": 20,
-    "eval_samples_per_second": 15.577,
-    "eval_steps_per_second": 2.337,
-    "total_flos": 775242448896.0,
-    "train_loss": 1.6423894166946411,
-    "train_runtime": 82.1401,
+    "epoch": 4.615384615384615,
+    "total_flos": 1161176285184.0,
+    "train_loss": 1.4103985627492268,
+    "train_runtime": 114.5958,
     "train_samples": 100,
-    "train_samples_per_second": 3.896,
-    "train_steps_per_second": 0.243
+    "train_samples_per_second": 4.189,
+    "train_steps_per_second": 0.262
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.076923076923077,
-    "total_flos": 775242448896.0,
-    "train_loss": 1.6423894166946411,
-    "train_runtime": 82.1401,
+    "epoch": 4.615384615384615,
+    "total_flos": 1161176285184.0,
+    "train_loss": 1.4103985627492268,
+    "train_runtime": 114.5958,
     "train_samples": 100,
-    "train_samples_per_second": 3.896,
-    "train_steps_per_second": 0.243
+    "train_samples_per_second": 4.189,
+    "train_steps_per_second": 0.262
 }

trainer_state.json

@@ -1,94 +1,124 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.076923076923077,
+  "epoch": 4.615384615384615,
   "eval_steps": 5,
-  "global_step": 20,
+  "global_step": 30,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.15384615384615385,
-      "grad_norm": 3.415924856190201,
-      "learning_rate": 0.0001,
+      "grad_norm": 3.415750820541017,
+      "learning_rate": 6.666666666666667e-05,
       "loss": 2.8513,
       "step": 1
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 1.3973726295486488,
-      "learning_rate": 0.00018660254037844388,
-      "loss": 2.3938,
+      "grad_norm": 1.0139669338401949,
+      "learning_rate": 0.00019730448705798239,
+      "loss": 2.4578,
       "step": 5
     },
     {
       "epoch": 0.7692307692307693,
-      "eval_loss": 1.758373498916626,
-      "eval_runtime": 2.4492,
-      "eval_samples_per_second": 8.166,
-      "eval_steps_per_second": 1.225,
+      "eval_loss": 1.7600507736206055,
+      "eval_runtime": 2.4719,
+      "eval_samples_per_second": 8.091,
+      "eval_steps_per_second": 1.214,
       "step": 5
     },
     {
       "epoch": 1.5384615384615383,
-      "grad_norm": 0.6743680553163675,
-      "learning_rate": 0.00011736481776669306,
-      "loss": 1.4879,
+      "grad_norm": 0.7815864493298474,
+      "learning_rate": 0.0001686241637868734,
+      "loss": 1.4794,
       "step": 10
     },
     {
       "epoch": 1.5384615384615383,
-      "eval_loss": 1.5969226360321045,
-      "eval_runtime": 1.2499,
-      "eval_samples_per_second": 16.001,
-      "eval_steps_per_second": 2.4,
+      "eval_loss": 1.5840351581573486,
+      "eval_runtime": 1.2327,
+      "eval_samples_per_second": 16.224,
+      "eval_steps_per_second": 2.434,
       "step": 10
     },
     {
       "epoch": 2.3076923076923075,
-      "grad_norm": 0.8800321123254317,
-      "learning_rate": 3.5721239031346066e-05,
-      "loss": 1.3838,
+      "grad_norm": 0.9857017227592326,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 1.3439,
       "step": 15
     },
     {
       "epoch": 2.3076923076923075,
-      "eval_loss": 1.544511079788208,
-      "eval_runtime": 1.2454,
-      "eval_samples_per_second": 16.06,
-      "eval_steps_per_second": 2.409,
+      "eval_loss": 1.4990818500518799,
+      "eval_runtime": 1.2487,
+      "eval_samples_per_second": 16.017,
+      "eval_steps_per_second": 2.403,
       "step": 15
     },
     {
       "epoch": 3.076923076923077,
-      "grad_norm": 0.8313117059613246,
-      "learning_rate": 0.0,
-      "loss": 1.2126,
+      "grad_norm": 0.992189655558923,
+      "learning_rate": 6.039202339608432e-05,
+      "loss": 1.0898,
       "step": 20
     },
     {
       "epoch": 3.076923076923077,
-      "eval_loss": 1.5321630239486694,
-      "eval_runtime": 1.2477,
-      "eval_samples_per_second": 16.029,
-      "eval_steps_per_second": 2.404,
+      "eval_loss": 1.4788802862167358,
+      "eval_runtime": 1.2331,
+      "eval_samples_per_second": 16.219,
+      "eval_steps_per_second": 2.433,
       "step": 20
     },
     {
-      "epoch": 3.076923076923077,
-      "step": 20,
-      "total_flos": 775242448896.0,
-      "train_loss": 1.6423894166946411,
-      "train_runtime": 82.1401,
-      "train_samples_per_second": 3.896,
-      "train_steps_per_second": 0.243
+      "epoch": 3.8461538461538463,
+      "grad_norm": 1.5484850409168303,
+      "learning_rate": 1.6451218858706374e-05,
+      "loss": 1.0223,
+      "step": 25
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "eval_loss": 1.509316086769104,
+      "eval_runtime": 1.2429,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.414,
+      "step": 25
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 1.3738605878063028,
+      "learning_rate": 0.0,
+      "loss": 0.9906,
+      "step": 30
+    },
+    {
+      "epoch": 4.615384615384615,
+      "eval_loss": 1.511298656463623,
+      "eval_runtime": 1.2383,
+      "eval_samples_per_second": 16.151,
+      "eval_steps_per_second": 2.423,
+      "step": 30
+    },
+    {
+      "epoch": 4.615384615384615,
+      "step": 30,
+      "total_flos": 1161176285184.0,
+      "train_loss": 1.4103985627492268,
+      "train_runtime": 114.5958,
+      "train_samples_per_second": 4.189,
+      "train_steps_per_second": 0.262
     }
   ],
   "logging_steps": 5,
-  "max_steps": 20,
+  "max_steps": 30,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 4,
+  "num_train_epochs": 5,
   "save_steps": 1000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -102,7 +132,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 775242448896.0,
+  "total_flos": 1161176285184.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null