Long-context reasoning on LongReason
86.9ScoreLONGTRACERL-GRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| LONGTRACERL-GRPOBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 86.9 | |
| DocQABackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 86.4 | |
| LONGTRACERLBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 85.4 | |
| LoongRLBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 84.9 | |
| BaseBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 84.2 | |
| LongRLVRBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 84.2 | |
| LONGTRACERLBackbone=Qwen3-4B-Thinking-25072026.05 | 83.8 | |
| LongRLVRBackbone=Qwen3-4B-Thinking-25072026.05 | 80.7 | |
| DocQABackbone=Qwen3-4B-Thinking-25072026.05 | 79.9 | |
| LoongRLBackbone=Qwen3-4B-Thinking-25072026.05 | 78.7 | |
| LONGTRACERL-GRPOBackbone=Qwen3-4B-Thinking-25072026.05 | 78.7 | |
| BaseBackbone=Qwen3-4B-Thinking-25072026.05 | 78.5 | |
| LONGTRACERL-GRPOBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 75.4 | |
| LONGTRACERLBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 75.2 | |
| BaseBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 74.1 | |
| DocQABackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 73.7 | |
| LoongRLBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 73.3 | |
| LongRLVRBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 73.2 |