Scene Decoding on CLEVR (test)
0Parse FailsNaive Baseline
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Naive BaselineCB=N/A2026.05 | 0 | 2.27 | 56.8 | 3.84 | 2.22 | |
| Qwen3.5-397B-A17BCB=322026.05 | 0 | 1.43 | 51 | 3.22 | 1.9 | |
| Qwen3.5-397B-A17BCB=642026.05 | 0 | 1 | 45.8 | 3.37 | 2 | |
| Qwen3.5-397B-A17BCB=1282026.05 | 0 | 1.53 | 50.5 | 3.17 | 1.91 | |
| DeepSeek-V3.1-vLLM-2CB=642026.05 | 0 | 0.93 | 54.3 | 3.64 | 1.97 | |
| DeepSeek-V3.1-vLLM-2CB=1282026.05 | 0 | 1.13 | 48.7 | 3.15 | 1.84 | |
| GPT-OSS-120BCB=322026.05 | 1 | 2.36 | 46.9 | 2.45 | 1.71 | |
| GPT-OSS-120BCB=1282026.05 | 1 | 2.36 | 52.4 | 4.61 | 2.13 | |
| DeepSeek-V3.1-vLLM-2CB=322026.05 | 1 | 1.36 | 51.3 | 3.68 | 2.11 | |
| GPT-OSS-120BCB=642026.05 | 3 | 2.83 | 53.6 | 3.26 | 1.87 |