End of training

README.md

@@ -1,17 +1,20 @@
 ---
 base_model: maxkordn/Qwen2.5-Coder-7B-Instruct-Solver-RFT
+datasets: flattened_successful_trajectories/v1
 library_name: transformers
 model_name: Qwen2.5-Coder-7B-Instruct-Solver-RFT
 tags:
 - generated_from_trainer
+- cogzero
 - trl
 - sft
+- kordn
 licence: license
 ---
 
 # Model Card for Qwen2.5-Coder-7B-Instruct-Solver-RFT
 
-This model is a fine-tuned version of [maxkordn/Qwen2.5-Coder-7B-Instruct-Solver-RFT](https://huggingface.co/maxkordn/Qwen2.5-Coder-7B-Instruct-Solver-RFT).
+This model is a fine-tuned version of [maxkordn/Qwen2.5-Coder-7B-Instruct-Solver-RFT](https://huggingface.co/maxkordn/Qwen2.5-Coder-7B-Instruct-Solver-RFT) on the [flattened_successful_trajectories/v1](https://huggingface.co/datasets/flattened_successful_trajectories/v1) dataset.
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start

all_results.json

@@ -1,4 +1,9 @@
 {
+    "eval_loss": 0.04499583691358566,
+    "eval_runtime": 57.7281,
+    "eval_samples": 2439,
+    "eval_samples_per_second": 42.25,
+    "eval_steps_per_second": 5.283,
     "total_flos": 9.471448716243108e+17,
     "train_loss": 0.017112477973285245,
     "train_runtime": 9496.6509,

config.json

@@ -52,7 +52,7 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.55.4",
-  "use_cache": false,
+  "use_cache": true,
   "use_sliding_window": false,
   "vocab_size": 152064
 }

eval_results.json

@@ -1,7 +1,7 @@
 {
-    "eval_loss": 0.10813906788825989,
-    "eval_runtime": 52.8222,
-    "eval_samples": 2505,
-    "eval_samples_per_second": 47.423,
-    "eval_steps_per_second": 5.944
+    "eval_loss": 0.04499583691358566,
+    "eval_runtime": 57.7281,
+    "eval_samples": 2439,
+    "eval_samples_per_second": 42.25,
+    "eval_steps_per_second": 5.283
 }