End of training

README.md

@@ -1,8 +1,13 @@
 ---
 base_model: barc0/Llama-3.1-ARC-Potpourri-Transduction-8B
+datasets:
+- tttx/problem0_data
+- barc0/transduction_formatted_rearc_dataset_100k
+- barc0/transduction_heavy_100k_jsonl
 library_name: peft
 license: llama3.1
 tags:
+- alignment-handbook
 - trl
 - sft
 - generated_from_trainer
@@ -16,7 +21,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # engineer1-heavy-barc-llama3.1-8b-instruct-lora64-testtime-finetuning
 
-This model is a fine-tuned version of [barc0/Llama-3.1-ARC-Potpourri-Transduction-8B](https://huggingface.co/barc0/Llama-3.1-ARC-Potpourri-Transduction-8B) on an unknown dataset.
+This model is a fine-tuned version of [barc0/Llama-3.1-ARC-Potpourri-Transduction-8B](https://huggingface.co/barc0/Llama-3.1-ARC-Potpourri-Transduction-8B) on the tttx/problem0_data, the barc0/transduction_formatted_rearc_dataset_100k and the barc0/transduction_heavy_100k_jsonl datasets.
 It achieves the following results on the evaluation set:
 - Loss: 0.0000
 

all_results.json

@@ -1,5 +1,10 @@
 {
     "epoch": 10.0,
+    "eval_loss": 2.355557808186859e-05,
+    "eval_runtime": 0.6765,
+    "eval_samples": 1,
+    "eval_samples_per_second": 1.478,
+    "eval_steps_per_second": 1.478,
     "total_flos": 4718728839168.0,
     "train_loss": 0.0003110202618358926,
     "train_runtime": 115.7621,

config.json

@@ -1,6 +1,5 @@
 {
-  "_attn_implementation_autoset": true,
-  "_name_or_path": "barc0/Llama-3.1-ARC-Heavy-Transduction-8B",
+  "_name_or_path": "barc0/Llama-3.1-ARC-Potpourri-Transduction-8B",
   "architectures": [
     "LlamaForCausalLM"
   ],
@@ -35,7 +34,7 @@
   "rope_theta": 500000.0,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.46.2",
+  "transformers_version": "4.45.2",
   "use_cache": true,
   "vocab_size": 128256
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.0,
-    "eval_loss": 0.09748542308807373,
-    "eval_runtime": 890.5512,
-    "eval_samples": 10376,
-    "eval_samples_per_second": 11.651,
-    "eval_steps_per_second": 1.456
+    "epoch": 10.0,
+    "eval_loss": 2.355557808186859e-05,
+    "eval_runtime": 0.6765,
+    "eval_samples": 1,
+    "eval_samples_per_second": 1.478,
+    "eval_steps_per_second": 1.478
 }