Model save

README.md

@@ -3,7 +3,6 @@ base_model: meta-llama/Llama-2-7b-chat-hf
 library_name: peft
 license: llama2
 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-2-7b-chat-hf](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.3313
+- Loss: 1.2229
 
 ## Model description
 
@@ -50,16 +49,22 @@ 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: 50
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 2.4105        | 0.7692 | 5    | 1.8901          |
-| 1.5392        | 1.5385 | 10   | 1.4715          |
-| 1.3211        | 2.3077 | 15   | 1.3548          |
-| 1.1911        | 3.0769 | 20   | 1.3313          |
+| 2.5966        | 0.7692 | 5    | 2.3082          |
+| 1.7267        | 1.5385 | 10   | 1.4897          |
+| 1.314         | 2.3077 | 15   | 1.3659          |
+| 1.1446        | 3.0769 | 20   | 1.2682          |
+| 1.0113        | 3.8462 | 25   | 1.2397          |
+| 0.9375        | 4.6154 | 30   | 1.2166          |
+| 0.9363        | 5.3846 | 35   | 1.2153          |
+| 0.8455        | 6.1538 | 40   | 1.2205          |
+| 0.8204        | 6.9231 | 45   | 1.2213          |
+| 0.8049        | 7.6923 | 50   | 1.2229          |
 
 
 ### Framework versions

all_results.json

@@ -1,14 +1,9 @@
 {
-    "epoch": 3.076923076923077,
-    "eval_loss": 1.3312898874282837,
-    "eval_runtime": 1.3952,
-    "eval_samples": 20,
-    "eval_samples_per_second": 14.334,
-    "eval_steps_per_second": 2.15,
-    "total_flos": 278393585664.0,
-    "train_loss": 1.6444377422332763,
-    "train_runtime": 82.8393,
+    "epoch": 7.6923076923076925,
+    "total_flos": 709396267008.0,
+    "train_loss": 1.22164155960083,
+    "train_runtime": 177.5633,
     "train_samples": 100,
-    "train_samples_per_second": 3.863,
-    "train_steps_per_second": 0.241
+    "train_samples_per_second": 4.505,
+    "train_steps_per_second": 0.282
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.076923076923077,
-    "total_flos": 278393585664.0,
-    "train_loss": 1.6444377422332763,
-    "train_runtime": 82.8393,
+    "epoch": 7.6923076923076925,
+    "total_flos": 709396267008.0,
+    "train_loss": 1.22164155960083,
+    "train_runtime": 177.5633,
     "train_samples": 100,
-    "train_samples_per_second": 3.863,
-    "train_steps_per_second": 0.241
+    "train_samples_per_second": 4.505,
+    "train_steps_per_second": 0.282
 }

trainer_state.json

@@ -1,94 +1,184 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.076923076923077,
+  "epoch": 7.6923076923076925,
   "eval_steps": 5,
-  "global_step": 20,
+  "global_step": 50,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.15384615384615385,
-      "grad_norm": 2.313300511918975,
-      "learning_rate": 0.0001,
+      "grad_norm": 2.313393041940284,
+      "learning_rate": 4e-05,
       "loss": 2.9898,
       "step": 1
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 1.519688558573064,
-      "learning_rate": 0.00018660254037844388,
-      "loss": 2.4105,
+      "grad_norm": 1.6130089228285052,
+      "learning_rate": 0.0002,
+      "loss": 2.5966,
       "step": 5
     },
     {
       "epoch": 0.7692307692307693,
-      "eval_loss": 1.8901169300079346,
-      "eval_runtime": 2.6055,
-      "eval_samples_per_second": 7.676,
-      "eval_steps_per_second": 1.151,
+      "eval_loss": 2.3082473278045654,
+      "eval_runtime": 2.6471,
+      "eval_samples_per_second": 7.555,
+      "eval_steps_per_second": 1.133,
       "step": 5
     },
     {
       "epoch": 1.5384615384615383,
-      "grad_norm": 0.5393161125298456,
-      "learning_rate": 0.00011736481776669306,
-      "loss": 1.5392,
+      "grad_norm": 0.5797187565008463,
+      "learning_rate": 0.00019396926207859084,
+      "loss": 1.7267,
       "step": 10
     },
     {
       "epoch": 1.5384615384615383,
-      "eval_loss": 1.4714559316635132,
-      "eval_runtime": 1.3474,
-      "eval_samples_per_second": 14.844,
-      "eval_steps_per_second": 2.227,
+      "eval_loss": 1.4897228479385376,
+      "eval_runtime": 1.3736,
+      "eval_samples_per_second": 14.56,
+      "eval_steps_per_second": 2.184,
       "step": 10
     },
     {
       "epoch": 2.3076923076923075,
-      "grad_norm": 1.0219537494358102,
-      "learning_rate": 3.5721239031346066e-05,
-      "loss": 1.3211,
+      "grad_norm": 1.547845061473479,
+      "learning_rate": 0.0001766044443118978,
+      "loss": 1.314,
       "step": 15
     },
     {
       "epoch": 2.3076923076923075,
-      "eval_loss": 1.3548352718353271,
-      "eval_runtime": 1.36,
-      "eval_samples_per_second": 14.706,
-      "eval_steps_per_second": 2.206,
+      "eval_loss": 1.3659324645996094,
+      "eval_runtime": 1.3743,
+      "eval_samples_per_second": 14.552,
+      "eval_steps_per_second": 2.183,
       "step": 15
     },
     {
       "epoch": 3.076923076923077,
-      "grad_norm": 0.4730950632176745,
-      "learning_rate": 0.0,
-      "loss": 1.1911,
+      "grad_norm": 0.5558495576003724,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.1446,
       "step": 20
     },
     {
       "epoch": 3.076923076923077,
-      "eval_loss": 1.3312898874282837,
-      "eval_runtime": 1.3488,
-      "eval_samples_per_second": 14.828,
-      "eval_steps_per_second": 2.224,
+      "eval_loss": 1.2681808471679688,
+      "eval_runtime": 1.3661,
+      "eval_samples_per_second": 14.64,
+      "eval_steps_per_second": 2.196,
       "step": 20
     },
     {
-      "epoch": 3.076923076923077,
-      "step": 20,
-      "total_flos": 278393585664.0,
-      "train_loss": 1.6444377422332763,
-      "train_runtime": 82.8393,
-      "train_samples_per_second": 3.863,
-      "train_steps_per_second": 0.241
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.7835871813274478,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 1.0113,
+      "step": 25
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "eval_loss": 1.2397050857543945,
+      "eval_runtime": 1.3601,
+      "eval_samples_per_second": 14.705,
+      "eval_steps_per_second": 2.206,
+      "step": 25
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.4313493686529163,
+      "learning_rate": 8.263518223330697e-05,
+      "loss": 0.9375,
+      "step": 30
+    },
+    {
+      "epoch": 4.615384615384615,
+      "eval_loss": 1.2165509462356567,
+      "eval_runtime": 1.3769,
+      "eval_samples_per_second": 14.525,
+      "eval_steps_per_second": 2.179,
+      "step": 30
+    },
+    {
+      "epoch": 5.384615384615385,
+      "grad_norm": 0.2940648413957461,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.9363,
+      "step": 35
+    },
+    {
+      "epoch": 5.384615384615385,
+      "eval_loss": 1.2152830362319946,
+      "eval_runtime": 1.3657,
+      "eval_samples_per_second": 14.645,
+      "eval_steps_per_second": 2.197,
+      "step": 35
+    },
+    {
+      "epoch": 6.153846153846154,
+      "grad_norm": 0.3584244840352673,
+      "learning_rate": 2.339555568810221e-05,
+      "loss": 0.8455,
+      "step": 40
+    },
+    {
+      "epoch": 6.153846153846154,
+      "eval_loss": 1.2205383777618408,
+      "eval_runtime": 1.3618,
+      "eval_samples_per_second": 14.686,
+      "eval_steps_per_second": 2.203,
+      "step": 40
+    },
+    {
+      "epoch": 6.923076923076923,
+      "grad_norm": 0.3722733679699954,
+      "learning_rate": 6.030737921409169e-06,
+      "loss": 0.8204,
+      "step": 45
+    },
+    {
+      "epoch": 6.923076923076923,
+      "eval_loss": 1.2212796211242676,
+      "eval_runtime": 1.367,
+      "eval_samples_per_second": 14.631,
+      "eval_steps_per_second": 2.195,
+      "step": 45
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "grad_norm": 0.29798153347655665,
+      "learning_rate": 0.0,
+      "loss": 0.8049,
+      "step": 50
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "eval_loss": 1.2228620052337646,
+      "eval_runtime": 1.3691,
+      "eval_samples_per_second": 14.608,
+      "eval_steps_per_second": 2.191,
+      "step": 50
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "step": 50,
+      "total_flos": 709396267008.0,
+      "train_loss": 1.22164155960083,
+      "train_runtime": 177.5633,
+      "train_samples_per_second": 4.505,
+      "train_steps_per_second": 0.282
     }
   ],
   "logging_steps": 5,
-  "max_steps": 20,
+  "max_steps": 50,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 4,
+  "num_train_epochs": 9,
   "save_steps": 1000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -102,7 +192,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 278393585664.0,
+  "total_flos": 709396267008.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null