Model save

README.md

@@ -1,16 +1,12 @@
 ---
 library_name: peft
 tags:
-- alignment-handbook
-- trl
-- sft
-- generated_from_trainer
 - trl
 - sft
 - alignment-handbook
 - generated_from_trainer
 datasets:
-- HuggingFaceH4/ultrachat_200k
+- generator
 base_model: mistralai/Mistral-7B-v0.1
 model-index:
 - name: zephyr-7b-pl-qlora
@@ -22,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-pl-qlora
 
-This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the HuggingFaceH4/ultrachat_200k dataset.
+This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the generator dataset.
 
 ## Model description
 

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 0.01,
-    "train_loss": 13996.267578125,
-    "train_runtime": 449.6404,
+    "train_loss": 1.0116421318054198,
+    "train_runtime": 473.1338,
     "train_samples": 207865,
-    "train_samples_per_second": 1.779,
-    "train_steps_per_second": 0.222
+    "train_samples_per_second": 1.691,
+    "train_steps_per_second": 0.211
 }

lora_0/adapter_config.json

@@ -23,13 +23,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "gate_proj",
-    "q_proj",
-    "v_proj",
+    "down_proj",
     "o_proj",
     "k_proj",
+    "gate_proj",
+    "v_proj",
     "up_proj",
-    "down_proj"
+    "q_proj"
   ],
   "task_type": null,
   "use_dora": false,

lora_0/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dba2e465325cced80a4478a6ecdb57fd17a47e36ceda7f92fd43bcd28a952cff
+oid sha256:ad8b98a44de7f55dc76067740df02dc07c1c3684a06d731190cc46be3c07c1b0
 size 167832240

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 0.01,
-    "train_loss": 13996.267578125,
-    "train_runtime": 449.6404,
+    "train_loss": 1.0116421318054198,
+    "train_runtime": 473.1338,
     "train_samples": 207865,
-    "train_samples_per_second": 1.779,
-    "train_steps_per_second": 0.222
+    "train_samples_per_second": 1.691,
+    "train_steps_per_second": 0.211
 }

trainer_state.json

@@ -10,32 +10,56 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 58005.6328,
+      "learning_rate": 2e-05,
+      "loss": 1.2234,
       "step": 1
     },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019396926207859084,
+      "loss": 1.1179,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.0056,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.263518223330697e-05,
+      "loss": 0.9795,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.339555568810221e-05,
+      "loss": 0.9954,
+      "step": 80
+    },
     {
       "epoch": 0.01,
-      "learning_rate": 1.339745962155613e-05,
-      "loss": 13551.7285,
+      "learning_rate": 0.0,
+      "loss": 0.9545,
       "step": 100
     },
     {
       "epoch": 0.01,
       "step": 100,
-      "total_flos": 7.072526552059085e+16,
-      "train_loss": 13996.267578125,
-      "train_runtime": 449.6404,
-      "train_samples_per_second": 1.779,
-      "train_steps_per_second": 0.222
+      "total_flos": 7.072526874181632e+16,
+      "train_loss": 1.0116421318054198,
+      "train_runtime": 473.1338,
+      "train_samples_per_second": 1.691,
+      "train_steps_per_second": 0.211
     }
   ],
-  "logging_steps": 100,
+  "logging_steps": 20,
   "max_steps": 100,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
-  "total_flos": 7.072526552059085e+16,
+  "total_flos": 7.072526874181632e+16,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null