Model save

README.md

@@ -2,16 +2,9 @@
 license: mit
 library_name: peft
 tags:
-- alignment-handbook
 - trl
 - sft
 - generated_from_trainer
-- trl
-- sft
-- generated_from_trainer
-datasets:
-- ChenWu98/skills_metaphor_chat
-- ChenWu98/skills_red_herring_chat
 base_model: HuggingFaceH4/zephyr-7b-beta
 model-index:
 - name: skills_metaphor_chat-skills_red_herring_chat-lora
@@ -23,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # skills_metaphor_chat-skills_red_herring_chat-lora
 
-This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta) on the ChenWu98/skills_metaphor_chat and the ChenWu98/skills_red_herring_chat datasets.
+This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.2247
+- Loss: 0.2123
 
 ## Model description
 
@@ -54,13 +47,14 @@ 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
-- num_epochs: 1.0
+- num_epochs: 2.0
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.203         | 0.99  | 37   | 0.2247          |
+| 0.199         | 0.99  | 37   | 0.2184          |
+| 0.1668        | 1.97  | 74   | 0.2123          |
 
 
 ### Framework versions

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 0.99,
-    "eval_loss": 0.22467635571956635,
-    "eval_runtime": 8.1549,
+    "epoch": 1.97,
+    "eval_loss": 0.21233995258808136,
+    "eval_runtime": 8.1232,
     "eval_samples": 200,
-    "eval_samples_per_second": 24.525,
-    "eval_steps_per_second": 3.066,
-    "train_loss": 0.5330296472923176,
-    "train_runtime": 203.4112,
+    "eval_samples_per_second": 24.621,
+    "eval_steps_per_second": 3.078,
+    "train_loss": 0.39267757534980774,
+    "train_runtime": 398.9638,
     "train_samples": 600,
-    "train_samples_per_second": 2.95,
-    "train_steps_per_second": 0.182
+    "train_samples_per_second": 3.008,
+    "train_steps_per_second": 0.185
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.99,
-    "eval_loss": 0.22467635571956635,
-    "eval_runtime": 8.1549,
+    "epoch": 1.97,
+    "eval_loss": 0.21233995258808136,
+    "eval_runtime": 8.1232,
     "eval_samples": 200,
-    "eval_samples_per_second": 24.525,
-    "eval_steps_per_second": 3.066
+    "eval_samples_per_second": 24.621,
+    "eval_steps_per_second": 3.078
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.99,
-    "train_loss": 0.5330296472923176,
-    "train_runtime": 203.4112,
+    "epoch": 1.97,
+    "train_loss": 0.39267757534980774,
+    "train_runtime": 398.9638,
     "train_samples": 600,
-    "train_samples_per_second": 2.95,
-    "train_steps_per_second": 0.182
+    "train_samples_per_second": 3.008,
+    "train_steps_per_second": 0.185
 }

trainer_state.json

@@ -1,85 +1,135 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9866666666666667,
+  "epoch": 1.9733333333333334,
   "eval_steps": 500,
-  "global_step": 37,
+  "global_step": 74,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.03,
-      "learning_rate": 5e-05,
+      "learning_rate": 2.5e-05,
       "loss": 2.2275,
       "step": 1
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00019954719225730847,
-      "loss": 1.9848,
+      "learning_rate": 0.000125,
+      "loss": 2.1443,
       "step": 5
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00018412535328311814,
-      "loss": 0.6529,
+      "learning_rate": 0.00019954719225730847,
+      "loss": 1.0226,
       "step": 10
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00015000000000000001,
-      "loss": 0.2662,
+      "learning_rate": 0.00019450008187146684,
+      "loss": 0.31,
       "step": 15
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00010475819158237425,
-      "loss": 0.238,
+      "learning_rate": 0.00018412535328311814,
+      "loss": 0.2456,
       "step": 20
     },
     {
       "epoch": 0.67,
-      "learning_rate": 5.845849869981137e-05,
-      "loss": 0.2332,
+      "learning_rate": 0.00016900790114821122,
+      "loss": 0.2362,
       "step": 25
     },
     {
       "epoch": 0.8,
-      "learning_rate": 2.139469052572127e-05,
-      "loss": 0.2287,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.2269,
       "step": 30
     },
     {
       "epoch": 0.93,
-      "learning_rate": 1.8071302737293295e-06,
-      "loss": 0.203,
+      "learning_rate": 0.00012817325568414297,
+      "loss": 0.199,
       "step": 35
     },
     {
       "epoch": 0.99,
-      "eval_loss": 0.22467635571956635,
-      "eval_runtime": 9.0559,
-      "eval_samples_per_second": 22.085,
-      "eval_steps_per_second": 2.761,
+      "eval_loss": 0.21835945546627045,
+      "eval_runtime": 8.9027,
+      "eval_samples_per_second": 22.465,
+      "eval_steps_per_second": 2.808,
       "step": 37
     },
     {
-      "epoch": 0.99,
-      "step": 37,
-      "total_flos": 20586009395200.0,
-      "train_loss": 0.5330296472923176,
-      "train_runtime": 203.4112,
-      "train_samples_per_second": 2.95,
-      "train_steps_per_second": 0.182
+      "epoch": 1.07,
+      "learning_rate": 0.00010475819158237425,
+      "loss": 0.1995,
+      "step": 40
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.107487556395901e-05,
+      "loss": 0.1853,
+      "step": 45
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.845849869981137e-05,
+      "loss": 0.1781,
+      "step": 50
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8184101377939476e-05,
+      "loss": 0.1821,
+      "step": 55
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.139469052572127e-05,
+      "loss": 0.1811,
+      "step": 60
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.036800464548157e-06,
+      "loss": 0.1728,
+      "step": 65
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8071302737293295e-06,
+      "loss": 0.1668,
+      "step": 70
+    },
+    {
+      "epoch": 1.97,
+      "eval_loss": 0.21233995258808136,
+      "eval_runtime": 8.0137,
+      "eval_samples_per_second": 24.957,
+      "eval_steps_per_second": 3.12,
+      "step": 74
+    },
+    {
+      "epoch": 1.97,
+      "step": 74,
+      "total_flos": 41279761350656.0,
+      "train_loss": 0.39267757534980774,
+      "train_runtime": 398.9638,
+      "train_samples_per_second": 3.008,
+      "train_steps_per_second": 0.185
     }
   ],
   "logging_steps": 5,
-  "max_steps": 37,
+  "max_steps": 74,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 2,
   "save_steps": 500,
-  "total_flos": 20586009395200.0,
+  "total_flos": 41279761350656.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null