| Add IFStruct v1.0 evaluation result | SaylorTwift | open | 0 | disc |
| Qwen3.5-4B brain atlas activation census, OV-circuits, and capability fence (32layers | juiceb0xc0de | open | 0 | disc |
| Multilingual powerhouse — testing for mobile deployment | 3morixd | open | 0 | disc |
| What's the difference between this and Qwen3.5-4B-Base? | KipTonic | closed | 0 | disc |
| Add EvalEval community eval results | EvalEvalBot | closed | 0 | disc |
| Add MMMU-Pro evaluation result | SaylorTwift | open | 0 | disc |
| Add ParseBench evaluation results | boyang-runllama | open | 0 | disc |
| Add ParseBench evaluation results | boyang-runllama | closed | 0 | disc |
| Add ParseBench evaluation results | boyang-runllama | closed | 0 | disc |
| fix chat template to avoid empty historical `<think>` blocks | latent-variable | open | 0 | disc |
| Can we have a FP8 version? | drguolai | open | 0 | disc |
| Add ScreenSpot-Pro evaluation result | merve | open | 0 | disc |
| Recipe for full tuning using trl? | celsowm | open | 0 | disc |
| is this genuinely just overfitting with brittleness pro max or what | unokayish182 | open | 0 | disc |
| Instruct | karouswissem | open | 0 | disc |
| Create generation_config.json | jalola | open | 0 | disc |
| QORA-4B is a 4-billion parameter language model with built-in vision. Pure Rust multimodal inference engine build on Qwen3.5-4B | drdraq | open | 0 | disc |
| jax | jax990 | draft | 0 | disc |
| openai.APIConnectionError: Connection error. | kfranic | open | 0 | disc |
| Coding | POISONX | open | 0 | disc |
| Installation Video and Testing - Step by Step | fahdmirzac | open | 0 | disc |
| MMLU EVAL DGX SPARK | RGMC98 | open | 0 | disc |
| Add MMLU-Pro evaluation result | SaylorTwift | open | 0 | disc |