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-debug
   results: []
@@ -15,7 +20,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-2b-gemma-dft-debug
 
-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: 5.7648
 

all_results.json

@@ -1,5 +1,10 @@
 {
     "epoch": 0.9981515711645101,
+    "eval_loss": 5.764816761016846,
+    "eval_runtime": 363.9913,
+    "eval_samples": 750,
+    "eval_samples_per_second": 2.668,
+    "eval_steps_per_second": 0.335,
     "total_flos": 5.260333472022528e+16,
     "train_loss": 7.369897298459653,
     "train_runtime": 2195.5957,

config.json

@@ -24,6 +24,6 @@
   "rope_theta": 10000.0,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.43.3",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 256000
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 0.9981515711645101,
-    "eval_loss": 5.75302791595459,
-    "eval_runtime": 422.4972,
+    "eval_loss": 5.764816761016846,
+    "eval_runtime": 363.9913,
     "eval_samples": 750,
-    "eval_samples_per_second": 2.298,
-    "eval_steps_per_second": 0.289
+    "eval_samples_per_second": 2.668,
+    "eval_steps_per_second": 0.335
 }

runs/Aug02_11-59-21_ale-distillm-8-0-0/events.out.tfevents.1722599098.ale-distillm-8-0-0.10851.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5953baedf1481273660c9a3aab7da30f5e9bd8a5467f95e180aa6745b446881
+size 359