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: 0.2123
+- Loss: 0.2245
 
 ## Model description
 
@@ -49,8 +42,8 @@ The following hyperparameters were used during training:
 - eval_batch_size: 8
 - seed: 42
 - distributed_type: multi-GPU
-- gradient_accumulation_steps: 4
-- total_train_batch_size: 16
+- gradient_accumulation_steps: 8
+- total_train_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
@@ -60,8 +53,8 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.199         | 0.99  | 37   | 0.2184          |
-| 0.1668        | 1.97  | 74   | 0.2123          |
+| 0.2788        | 0.96  | 18   | 0.2390          |
+| 0.1993        | 1.92  | 36   | 0.2245          |
 
 
 ### Framework versions

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 1.97,
-    "eval_loss": 0.21233995258808136,
-    "eval_runtime": 8.1232,
+    "epoch": 1.92,
+    "eval_loss": 0.2245262861251831,
+    "eval_runtime": 10.1729,
     "eval_samples": 200,
-    "eval_samples_per_second": 24.621,
-    "eval_steps_per_second": 3.078,
-    "train_loss": 0.39267757534980774,
-    "train_runtime": 398.9638,
+    "eval_samples_per_second": 19.66,
+    "eval_steps_per_second": 2.458,
+    "train_loss": 0.5313632095025645,
+    "train_runtime": 554.6126,
     "train_samples": 600,
-    "train_samples_per_second": 3.008,
-    "train_steps_per_second": 0.185
+    "train_samples_per_second": 2.164,
+    "train_steps_per_second": 0.065
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.97,
-    "eval_loss": 0.21233995258808136,
-    "eval_runtime": 8.1232,
+    "epoch": 1.92,
+    "eval_loss": 0.2245262861251831,
+    "eval_runtime": 10.1729,
     "eval_samples": 200,
-    "eval_samples_per_second": 24.621,
-    "eval_steps_per_second": 3.078
+    "eval_samples_per_second": 19.66,
+    "eval_steps_per_second": 2.458
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.97,
-    "train_loss": 0.39267757534980774,
-    "train_runtime": 398.9638,
+    "epoch": 1.92,
+    "train_loss": 0.5313632095025645,
+    "train_runtime": 554.6126,
     "train_samples": 600,
-    "train_samples_per_second": 3.008,
-    "train_steps_per_second": 0.185
+    "train_samples_per_second": 2.164,
+    "train_steps_per_second": 0.065
 }

trainer_state.json

@@ -1,135 +1,93 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.9733333333333334,
+  "epoch": 1.92,
   "eval_steps": 500,
-  "global_step": 74,
+  "global_step": 36,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.03,
-      "learning_rate": 2.5e-05,
-      "loss": 2.2275,
+      "epoch": 0.05,
+      "learning_rate": 5e-05,
+      "loss": 2.272,
       "step": 1
     },
-    {
-      "epoch": 0.13,
-      "learning_rate": 0.000125,
-      "loss": 2.1443,
-      "step": 5
-    },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00019954719225730847,
-      "loss": 1.0226,
-      "step": 10
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 0.00019450008187146684,
-      "loss": 0.31,
-      "step": 15
+      "learning_rate": 0.0001995184726672197,
+      "loss": 1.954,
+      "step": 5
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00018412535328311814,
-      "loss": 0.2456,
-      "step": 20
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00016900790114821122,
-      "loss": 0.2362,
-      "step": 25
+      "learning_rate": 0.00018314696123025454,
+      "loss": 0.6371,
+      "step": 10
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00015000000000000001,
-      "loss": 0.2269,
-      "step": 30
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00012817325568414297,
-      "loss": 0.199,
-      "step": 35
+      "learning_rate": 0.0001471396736825998,
+      "loss": 0.2788,
+      "step": 15
     },
     {
-      "epoch": 0.99,
-      "eval_loss": 0.21835945546627045,
-      "eval_runtime": 8.9027,
-      "eval_samples_per_second": 22.465,
-      "eval_steps_per_second": 2.808,
-      "step": 37
+      "epoch": 0.96,
+      "eval_loss": 0.23904697597026825,
+      "eval_runtime": 11.1391,
+      "eval_samples_per_second": 17.955,
+      "eval_steps_per_second": 2.244,
+      "step": 18
     },
     {
       "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
+      "learning_rate": 0.0001,
+      "loss": 0.2252,
+      "step": 20
     },
     {
       "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
+      "learning_rate": 5.286032631740023e-05,
+      "loss": 0.2128,
+      "step": 25
     },
     {
       "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
+      "learning_rate": 1.6853038769745467e-05,
+      "loss": 0.2122,
+      "step": 30
     },
     {
       "epoch": 1.87,
-      "learning_rate": 1.8071302737293295e-06,
-      "loss": 0.1668,
-      "step": 70
+      "learning_rate": 4.815273327803182e-07,
+      "loss": 0.1993,
+      "step": 35
     },
     {
-      "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": 1.92,
+      "eval_loss": 0.2245262861251831,
+      "eval_runtime": 9.6856,
+      "eval_samples_per_second": 20.649,
+      "eval_steps_per_second": 2.581,
+      "step": 36
     },
     {
-      "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
+      "epoch": 1.92,
+      "step": 36,
+      "total_flos": 40159503319040.0,
+      "train_loss": 0.5313632095025645,
+      "train_runtime": 554.6126,
+      "train_samples_per_second": 2.164,
+      "train_steps_per_second": 0.065
     }
   ],
   "logging_steps": 5,
-  "max_steps": 74,
+  "max_steps": 36,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 2,
   "save_steps": 500,
-  "total_flos": 41279761350656.0,
+  "total_flos": 40159503319040.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null