End of training

README.md

@@ -1,10 +1,15 @@
 ---
+base_model: data/gemma-2b
 tags:
+- alignment-handbook
+- trl
+- sft
+- generated_from_trainer
 - trl
 - sft
 - generated_from_trainer
 datasets:
-- generator
+- argilla/dpo-mix-7k
 model-index:
 - name: zephyr-2b-gemma-dft
   results: []
@@ -16,7 +21,9 @@ should probably proofread and complete it, then remove this comment. -->
 [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://zebra.wandb.io/cto/distillm/runs/vxz2pnbq)
 # zephyr-2b-gemma-dft
 
-This model was trained from scratch on the generator dataset.
+This model is a fine-tuned version of [data/gemma-2b](https://huggingface.co/data/gemma-2b) on the argilla/dpo-mix-7k dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.0001
 
 ## Model description
 

all_results.json

@@ -1,5 +1,10 @@
 {
     "epoch": 0,
+    "eval_loss": 5.277253148960881e-05,
+    "eval_runtime": 76.218,
+    "eval_samples": 750,
+    "eval_samples_per_second": 12.74,
+    "eval_steps_per_second": 1.601,
     "total_flos": 0,
     "train_loss": 0.0,
     "train_runtime": 5.1094,

config.json

@@ -24,6 +24,6 @@
   "rope_theta": 10000.0,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.42.4",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 256000
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 0,
     "eval_loss": 5.277253148960881e-05,
-    "eval_runtime": 76.5403,
+    "eval_runtime": 76.218,
     "eval_samples": 750,
-    "eval_samples_per_second": 12.686,
-    "eval_steps_per_second": 1.594
+    "eval_samples_per_second": 12.74,
+    "eval_steps_per_second": 1.601
 }

runs/Jul23_16-11-48_ale-distillm-8-0-0/events.out.tfevents.1721747639.ale-distillm-8-0-0.69762.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bebad7a947c42b08bd748f2d0e8f90adef828c9a03e02faecdf0737fef4f3af1
+size 344