Feat/curr adj (#394)

training/evaluations/eval_algorithmic_composite.yaml

@@ -0,0 +1,28 @@
+# Model configuration
+model_path: ../utils/qwen3b_500 # Change to the smaller model
+max_tokens: 1024  # From max_response_length in training config
+temperature: 0.7  # Lower temperature for more focused responses
+top_p: 0.9  # From rollout top_p
+developer_prompt: DeepSeekZero
+developer_role: system  # Standard role for system prompts
+
+# Output configuration
+output_dir: results
+save_metadata: true
+save_full_results: true
+eval_repeats: 3
+
+# Categories and datasets to evaluate
+categories:
+  - category: reasoning
+    datasets:
+      - dataset: number_sorting
+        size: 100
+        seed: 42
+        params:
+          min_numbers: 3
+          max_numbers: 10
+          min_decimals: 0
+          max_decimals: 2
+          min_value: -100.0
+          max_value: 100.0