<feat>

logs/results_mathvision/20260810_105351_results.json

@@ -0,0 +1,203 @@
+{
+  "results": {
+    "mathvision_test": {
+      "alias": "mathvision_test",
+      "mathvision_standard_eval,none": 33.12,
+      "mathvision_standard_eval_stderr,none": "N/A",
+      "mathvision_standard_eval_stderr_clt,none": "N/A",
+      "mathvision_standard_eval_stderr_clustered,none": "N/A"
+    }
+  },
+  "group_subtasks": {
+    "mathvision_test": []
+  },
+  "configs": {
+    "mathvision_test": {
+      "task": "mathvision_test",
+      "dataset_path": "MathLLMs/MathVision",
+      "dataset_kwargs": {
+        "token": true
+      },
+      "test_split": "test",
+      "full_docs": false,
+      "process_results_use_image": false,
+      "doc_to_visual": "<function mathvision_doc_to_visual at 0x7ff720a4aef0>",
+      "doc_to_text": "<function mathvision_doc_to_text at 0x7ff74cee40d0>",
+      "doc_to_target": "answer",
+      "process_results": "<function mathvision_process_results at 0x7ff75cce1d80>",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "mathvision_standard_eval",
+          "aggregation": "<function mathvision_aggregate_results_eval at 0x7ff721af2680>",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "max_new_tokens": 16384,
+        "until": [
+          "\n\n"
+        ]
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "score_key": "score",
+      "lmms_eval_specific_kwargs": {
+        "default": {
+          "prompt_format": "solution",
+          "mc_prompt": "Answer the question with the option's letter from the given choices directly.",
+          "short_answer_prompt": "Answer the question with a number directly."
+        },
+        "prompt_format": "solution",
+        "mc_prompt": "Answer the question with the option's letter from the given choices directly.",
+        "short_answer_prompt": "Answer the question with a number directly."
+      },
+      "auto_strip_thinking": false
+    }
+  },
+  "versions": {
+    "mathvision_test": "Yaml"
+  },
+  "n-shot": {
+    "mathvision_test": 0
+  },
+  "higher_is_better": {
+    "mathvision_test": {
+      "mathvision_standard_eval": true
+    }
+  },
+  "n-samples": {
+    "mathvision_test": {
+      "original": 3040,
+      "effective": 3040
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": "model=/project/visgroup/ycaobd/backup_huawei/pretrained_models/retrained_on_nvidia/recons_baseline_cotv3_2_threshold_03_without_pretrain_no_geometry_lr1f6_gn_5_100step_bs8,tensor_parallel_size=8,gpu_memory_utilization=0.85,hf_overrides={\"architectures\":[\"Qwen3VLForConditionalGeneration\"]}",
+    "model_backend": "lmms_eval.models.chat.vllm.VLLM (chat)",
+    "batch_size": "8",
+    "batch_sizes": [],
+    "device": null,
+    "use_cache": null,
+    "limit": null,
+    "offset": 0,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": "",
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234,
+    "resolved_cli_args": {
+      "config": "",
+      "model": "vllm",
+      "tasks": "mathvision_test",
+      "model_args": "model=/project/visgroup/ycaobd/backup_huawei/pretrained_models/retrained_on_nvidia/recons_baseline_cotv3_2_threshold_03_without_pretrain_no_geometry_lr1f6_gn_5_100step_bs8,tensor_parallel_size=8,gpu_memory_utilization=0.85,hf_overrides={\"architectures\":[\"Qwen3VLForConditionalGeneration\"]}",
+      "launcher_args": null,
+      "num_fewshot": null,
+      "batch_size": "8",
+      "max_batch_size": null,
+      "device": null,
+      "output_path": "/project/visgroup/ycaobd/backup_huawei/exp_nvidia/eval/logs/retrain_on_nvidia_gpu/recons_baseline_cotv3_2_threshold_03_without_pretrain_no_geometry_lr1f6_gn_5_100step_bs8/20260810_105322/mathvision_lmms_eval_vllm",
+      "limit": null,
+      "offset": 0,
+      "use_cache": null,
+      "cache_requests": null,
+      "check_integrity": false,
+      "write_out": false,
+      "log_samples": true,
+      "wandb_log_samples": false,
+      "swanlab_log_samples": false,
+      "log_samples_suffix": "cotv3_2_threshold_03_without_pretrain_no_geometry_lr1f6_gn_5_100step_bs8_mathvision_vllm",
+      "system_instruction": null,
+      "apply_chat_template": false,
+      "fewshot_as_multiturn": false,
+      "show_config": false,
+      "include_path": null,
+      "gen_kwargs": "",
+      "reasoning_tags": "[[\"<think>\", \"</think>\"], [\"<analysis>\", \"</analysis>\"]]",
+      "verbosity": "DEBUG",
+      "wandb_args": "",
+      "swanlab_args": "",
+      "timezone": "Asia/Singapore",
+      "hf_hub_log_args": ",output_path=/project/visgroup/ycaobd/backup_huawei/exp_nvidia/eval/logs/retrain_on_nvidia_gpu/recons_baseline_cotv3_2_threshold_03_without_pretrain_no_geometry_lr1f6_gn_5_100step_bs8/20260810_105322/mathvision_lmms_eval_vllm",
+      "predict_only": false,
+      "seed": [
+        0,
+        1234,
+        1234,
+        1234
+      ],
+      "trust_remote_code": false,
+      "process_with_media": false,
+      "agentic_trace_mode": "basic",
+      "force_simple": false,
+      "tui": false,
+      "repeats": 1,
+      "baseline": null,
+      "max_tokens": null,
+      "power_analysis": false,
+      "effect_size": 0.03,
+      "alpha": 0.05,
+      "power": 0.8,
+      "correlation": 0.5,
+      "std_a": null,
+      "std_b": null
+    }
+  },
+  "git_hash": "v0.6-181-g75571cfe",
+  "git_branch": "main",
+  "lmms_eval_version": "main@v0.6-181",
+  "date": "20260810_105351",
+  "throughput": {
+    "total_gen_tokens": 30663060,
+    "total_elapsed_time": 16648.094818353653,
+    "avg_speed": 1841.835977903945,
+    "avg_ttft": 0.11515913190026032,
+    "avg_tpot": 0.0026323851096919223,
+    "avg_rank": 0.0
+  },
+  "usage": {},
+  "efficiency": {
+    "by_task": {
+      "mathvision_test": {
+        "docs": 3040.0,
+        "docs_with_token_counts": 0.0,
+        "total_input_tokens": 0.0,
+        "total_output_tokens": 0.0,
+        "total_tokens": 0.0,
+        "total_correct_score": 0.0,
+        "avg_output_tokens_per_sample": 0.0,
+        "tokens_per_correct_answer": null
+      }
+    },
+    "overall": {
+      "docs": 3040.0,
+      "docs_with_token_counts": 0.0,
+      "total_input_tokens": 0.0,
+      "total_output_tokens": 0.0,
+      "total_tokens": 0.0,
+      "total_correct_score": 0.0,
+      "avg_output_tokens_per_sample": 0.0,
+      "tokens_per_correct_answer": null
+    }
+  },
+  "task_hashes": {
+    "mathvision_test": "0d773ff7fa3851f0614c4e47855581cb3d51947d4e8f435580fb29aa01b582f4"
+  },
+  "model_source": "vllm",
+  "model_name": "/project/visgroup/ycaobd/backup_huawei/pretrained_models/retrained_on_nvidia/recons_baseline_cotv3_2_threshold_03_without_pretrain_no_geometry_lr1f6_gn_5_100step_bs8",
+  "model_name_sanitized": "retrained_on_nvidia__recons_baseline_cotv3_2_threshold_03_without_pretrain_no_geometry_lr1f6_gn_5_100step_bs8",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": false,
+  "chat_template": null,
+  "chat_template_sha": null,
+  "start_time": 392054.104369715,
+  "end_time": 408930.176954995,
+  "total_evaluation_time_seconds": "16876.072585280053"
+}

logs/results_mme/mme_metrics.json

@@ -0,0 +1,21 @@
+{
+  "by_category": {
+    "code_reasoning": 155.0,
+    "artwork": 144.5,
+    "celebrity": 163.8235294117647,
+    "numerical_calculation": 140.0,
+    "text_translation": 192.5,
+    "count": 163.33333333333334,
+    "color": 190.0,
+    "commonsense_reasoning": 152.85714285714286,
+    "position": 173.33333333333334,
+    "OCR": 177.5,
+    "landmark": 175.5,
+    "scene": 155.75,
+    "existence": 190.0,
+    "posters": 152.38095238095238
+  },
+  "perception_score": 1686.1211484593837,
+  "cognition_score": 640.3571428571429,
+  "total_score": 2326.4782913165263
+}

logs/results_vsibench/vsibench_metrics.json

@@ -0,0 +1,11 @@
+{
+  "object_rel_direction_accuracy": 0.4667421217192038,
+  "object_rel_distance_accuracy": 0.6098591549295774,
+  "route_planning_accuracy": 0.32989690721649484,
+  "obj_appearance_order_accuracy": 0.7168284789644013,
+  "object_abs_distance_MRA:.5:.95:.05": 0.49652278177458037,
+  "object_counting_MRA:.5:.95:.05": 0.6745132743362833,
+  "object_size_estimation_MRA:.5:.95:.05": 0.769884575026233,
+  "room_size_estimation_MRA:.5:.95:.05": 0.5479166666666667,
+  "overall": 0.57652049507918
+}