Model save

README.md

@@ -2,15 +2,9 @@
 license: mit
 library_name: peft
 tags:
-- alignment-handbook
 - trl
 - sft
 - generated_from_trainer
-- trl
-- sft
-- generated_from_trainer
-datasets:
-- ChenWu98/skills_red_herring_metaphor_chat
 base_model: HuggingFaceH4/zephyr-7b-beta
 model-index:
 - name: skills_red_herring_metaphor_chat-lora
@@ -22,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # skills_red_herring_metaphor_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_red_herring_metaphor_chat dataset.
+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: 0.3462
+- Loss: 0.3410
 
 ## Model description
 
@@ -48,18 +42,19 @@ 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
-- num_epochs: 1.0
+- num_epochs: 2.0
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.3652        | 0.96  | 18   | 0.3462          |
+| 1.6966        | 0.96  | 9    | 0.4003          |
+| 0.3419        | 1.92  | 18   | 0.3410          |
 
 
 ### Framework versions

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 0.96,
-    "eval_loss": 0.3462468981742859,
-    "eval_runtime": 4.2622,
+    "epoch": 1.92,
+    "eval_loss": 0.3410404324531555,
+    "eval_runtime": 5.319,
     "eval_samples": 100,
-    "eval_samples_per_second": 23.462,
-    "eval_steps_per_second": 3.05,
-    "train_loss": 0.8159759442011515,
-    "train_runtime": 112.608,
+    "eval_samples_per_second": 18.8,
+    "eval_steps_per_second": 2.444,
+    "train_loss": 0.8035296764638689,
+    "train_runtime": 289.9665,
     "train_samples": 300,
-    "train_samples_per_second": 2.664,
-    "train_steps_per_second": 0.16
+    "train_samples_per_second": 2.069,
+    "train_steps_per_second": 0.062
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.96,
-    "eval_loss": 0.3462468981742859,
-    "eval_runtime": 4.2622,
+    "epoch": 1.92,
+    "eval_loss": 0.3410404324531555,
+    "eval_runtime": 5.319,
     "eval_samples": 100,
-    "eval_samples_per_second": 23.462,
-    "eval_steps_per_second": 3.05
+    "eval_samples_per_second": 18.8,
+    "eval_steps_per_second": 2.444
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.96,
-    "train_loss": 0.8159759442011515,
-    "train_runtime": 112.608,
+    "epoch": 1.92,
+    "train_loss": 0.8035296764638689,
+    "train_runtime": 289.9665,
     "train_samples": 300,
-    "train_samples_per_second": 2.664,
-    "train_steps_per_second": 0.16
+    "train_samples_per_second": 2.069,
+    "train_steps_per_second": 0.062
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.96,
+  "epoch": 1.92,
   "eval_steps": 500,
   "global_step": 18,
   "is_hyper_param_search": false,
@@ -9,53 +9,61 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 0.0001,
-      "loss": 2.1555,
+      "loss": 2.1556,
       "step": 1
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.53,
       "learning_rate": 0.00018314696123025454,
-      "loss": 1.69,
+      "loss": 1.6966,
       "step": 5
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.96,
+      "eval_loss": 0.4002748727798462,
+      "eval_runtime": 5.8667,
+      "eval_samples_per_second": 17.045,
+      "eval_steps_per_second": 2.216,
+      "step": 9
+    },
+    {
+      "epoch": 1.07,
       "learning_rate": 0.0001,
-      "loss": 0.5879,
+      "loss": 0.5695,
       "step": 10
     },
     {
-      "epoch": 0.8,
+      "epoch": 1.6,
       "learning_rate": 1.6853038769745467e-05,
-      "loss": 0.3652,
+      "loss": 0.3419,
       "step": 15
     },
     {
-      "epoch": 0.96,
-      "eval_loss": 0.3462468981742859,
-      "eval_runtime": 5.1291,
-      "eval_samples_per_second": 19.496,
-      "eval_steps_per_second": 2.535,
+      "epoch": 1.92,
+      "eval_loss": 0.3410404324531555,
+      "eval_runtime": 7.4319,
+      "eval_samples_per_second": 13.456,
+      "eval_steps_per_second": 1.749,
       "step": 18
     },
     {
-      "epoch": 0.96,
+      "epoch": 1.92,
       "step": 18,
-      "total_flos": 13668439425024.0,
-      "train_loss": 0.8159759442011515,
-      "train_runtime": 112.608,
-      "train_samples_per_second": 2.664,
-      "train_steps_per_second": 0.16
+      "total_flos": 27578625622016.0,
+      "train_loss": 0.8035296764638689,
+      "train_runtime": 289.9665,
+      "train_samples_per_second": 2.069,
+      "train_steps_per_second": 0.062
     }
   ],
   "logging_steps": 5,
   "max_steps": 18,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 2,
   "save_steps": 500,
-  "total_flos": 13668439425024.0,
+  "total_flos": 27578625622016.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null