Model save

README.md

@@ -1,16 +1,12 @@
 ---
-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
@@ -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
 
@@ -46,6 +42,9 @@ The following hyperparameters were used during training:
 - eval_batch_size: 4
 - seed: 42
 - distributed_type: multi-GPU
+- num_devices: 4
+- total_train_batch_size: 8
+- total_eval_batch_size: 16
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1

adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "q_proj",
-    "up_proj",
-    "v_proj",
+    "down_proj",
+    "o_proj",
     "k_proj",
+    "q_proj",
     "gate_proj",
-    "o_proj",
-    "down_proj"
+    "up_proj",
+    "v_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.05,
-    "train_loss": 0.0,
-    "train_runtime": 10.3365,
+    "epoch": 0.21,
+    "train_loss": 15.370767613482842,
+    "train_runtime": 10.0734,
     "train_samples": 207865,
-    "train_samples_per_second": 0.193,
-    "train_steps_per_second": 0.097
+    "train_samples_per_second": 0.794,
+    "train_steps_per_second": 0.099
 }

lora_0/adapter_config.json

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

runs/Apr18_10-03-34_g3013/events.out.tfevents.1713459912.g3013.93731.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65407f99bceee79f9ed49c49176eadcbd0e2e28fb2d85c74fad77a7c78406f06
+size 4803

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.05,
-    "train_loss": 0.0,
-    "train_runtime": 10.3365,
+    "epoch": 0.21,
+    "train_loss": 15.370767613482842,
+    "train_runtime": 10.0734,
     "train_samples": 207865,
-    "train_samples_per_second": 0.193,
-    "train_steps_per_second": 0.097
+    "train_samples_per_second": 0.794,
+    "train_steps_per_second": 0.099
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.05308909385623915,
+  "epoch": 0.21234723736301567,
   "eval_steps": 500,
   "global_step": 3701,
   "is_hyper_param_search": false,
@@ -4455,13 +4455,13 @@
       "step": 3700
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.21,
       "step": 3701,
-      "total_flos": 2.617011781310087e+18,
-      "train_loss": 0.0,
-      "train_runtime": 10.3365,
-      "train_samples_per_second": 0.193,
-      "train_steps_per_second": 0.097
+      "total_flos": 2.6175422208105513e+18,
+      "train_loss": 15.370767613482842,
+      "train_runtime": 10.0734,
+      "train_samples_per_second": 0.794,
+      "train_steps_per_second": 0.099
     }
   ],
   "logging_steps": 5,
@@ -4469,7 +4469,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
-  "total_flos": 2.617011781310087e+18,
+  "total_flos": 2.6175422208105513e+18,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1d9a715f7a325491aa2c19eea5014b4f41b6c6737c32607791c80dc00a8932b6
+oid sha256:54fc1c91b0b5d4934d9667747136e5185061d638e0ef1d7fef6ae4ae1d94f103
 size 4792