| { |
| "checkpoint_load": { |
| "mae": { |
| "matched_tensor_keys": 142, |
| "missing_locpath_weights": 0, |
| "unexpected_weights": 0, |
| "loaded_values_match_export": true, |
| "query_shape": [ |
| 128, |
| 512 |
| ] |
| }, |
| "stage1": { |
| "matched_tensor_keys": 148, |
| "missing_locpath_weights": 0, |
| "unexpected_weights": 0, |
| "loaded_values_match_export": true, |
| "query_shape": [ |
| 128, |
| 512 |
| ] |
| }, |
| "stage2": { |
| "matched_tensor_keys": 540, |
| "missing_locpath_weights": 0, |
| "unexpected_weights": 0, |
| "loaded_values_match_export": true, |
| "query_shape": [ |
| 128, |
| 512 |
| ] |
| } |
| }, |
| "tensor_export": { |
| "finite": true, |
| "tensor_only_roundtrip": true, |
| "stored_dtype": "float16", |
| "resampler_query_shape": [ |
| 128, |
| 512 |
| ] |
| }, |
| "trained_model_inference": { |
| "predictions": 2, |
| "tasks": [ |
| "1_Report_test.json", |
| "bcnb_choice_test.json" |
| ], |
| "all_outputs_nonempty": true, |
| "base_model": "Qwen/Qwen2.5-7B-Instruct", |
| "torch_dtype": "bfloat16", |
| "quantization": "NF4", |
| "feature_sampling_limit": 1024, |
| "resampler_latents": 128, |
| "router_top_m": 96, |
| "max_new_tokens": 96, |
| "full_benchmark_rerun": false |
| } |
| } |
|
|
