Model save

README.md

@@ -2,16 +2,12 @@
 license: apache-2.0
 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
@@ -23,7 +19,7 @@ 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
 
@@ -42,11 +38,13 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.0002
+- learning_rate: 5e-07
 - train_batch_size: 2
 - eval_batch_size: 4
 - seed: 42
 - distributed_type: multi-GPU
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 8
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.0,
-    "train_loss": 0.0,
-    "train_runtime": 501.823,
+    "epoch": 0.01,
+    "train_loss": 0.6359813857078552,
+    "train_runtime": 2120.9698,
     "train_samples": 207865,
-    "train_samples_per_second": 0.399,
-    "train_steps_per_second": 0.199
+    "train_samples_per_second": 0.377,
+    "train_steps_per_second": 0.047
 }

lora_0/adapter_config.json

@@ -23,13 +23,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
-    "o_proj",
-    "k_proj",
     "gate_proj",
-    "v_proj",
     "up_proj",
-    "q_proj"
+    "q_proj",
+    "v_proj",
+    "o_proj",
+    "k_proj",
+    "down_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:ad8b98a44de7f55dc76067740df02dc07c1c3684a06d731190cc46be3c07c1b0
+oid sha256:d95350f5041ab4e9166125ad0900a94e5edbb0d39d4fdfef11b30e664c3a0d10
 size 167832240

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.0,
-    "train_loss": 0.0,
-    "train_runtime": 501.823,
+    "epoch": 0.01,
+    "train_loss": 0.6359813857078552,
+    "train_runtime": 2120.9698,
     "train_samples": 207865,
-    "train_samples_per_second": 0.399,
-    "train_steps_per_second": 0.199
+    "train_samples_per_second": 0.377,
+    "train_steps_per_second": 0.047
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.0014344526845781992,
+  "epoch": 0.005737810738312797,
   "eval_steps": 500,
   "global_step": 100,
   "is_hyper_param_search": false,
@@ -10,56 +10,146 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 2e-05,
-      "loss": 0.0,
+      "learning_rate": 5e-08,
+      "loss": 0.6931,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.00019396926207859084,
-      "loss": 0.0,
+      "learning_rate": 2.5e-07,
+      "loss": 0.6931,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-07,
+      "loss": 0.6921,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.96201938253052e-07,
+      "loss": 0.6886,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.849231551964771e-07,
+      "loss": 0.6827,
       "step": 20
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.00015000000000000001,
-      "loss": 0.0,
+      "learning_rate": 4.6650635094610966e-07,
+      "loss": 0.6745,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.415111107797445e-07,
+      "loss": 0.6645,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.106969024216348e-07,
+      "loss": 0.6549,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.75e-07,
+      "loss": 0.6463,
       "step": 40
     },
     {
       "epoch": 0.0,
-      "learning_rate": 8.263518223330697e-05,
-      "loss": 0.0,
+      "learning_rate": 3.355050358314172e-07,
+      "loss": 0.6395,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.934120444167326e-07,
+      "loss": 0.6305,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-07,
+      "loss": 0.6241,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.065879555832674e-07,
+      "loss": 0.6155,
       "step": 60
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.339555568810221e-05,
-      "loss": 0.0,
+      "learning_rate": 1.6449496416858282e-07,
+      "loss": 0.6147,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2500000000000005e-07,
+      "loss": 0.6042,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.930309757836516e-08,
+      "loss": 0.6041,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.848888922025552e-08,
+      "loss": 0.5989,
       "step": 80
     },
     {
       "epoch": 0.0,
+      "learning_rate": 3.349364905389032e-08,
+      "loss": 0.5988,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.507684480352292e-08,
+      "loss": 0.5961,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.798061746947995e-09,
+      "loss": 0.5984,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 0.0,
+      "loss": 0.5982,
       "step": 100
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "step": 100,
-      "total_flos": 1.768131718545408e+16,
-      "train_loss": 0.0,
-      "train_runtime": 501.823,
-      "train_samples_per_second": 0.399,
-      "train_steps_per_second": 0.199
+      "total_flos": 7.072526927868723e+16,
+      "train_loss": 0.6359813857078552,
+      "train_runtime": 2120.9698,
+      "train_samples_per_second": 0.377,
+      "train_steps_per_second": 0.047
     }
   ],
-  "logging_steps": 20,
+  "logging_steps": 5,
   "max_steps": 100,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
-  "total_flos": 1.768131718545408e+16,
+  "total_flos": 7.072526927868723e+16,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null