Search results

Question: temperature/top_p for training data generation

Status: Open.

#33 In z-lab/dflash;

· ycchen-tw opened on Mar 2, 2026

qwen3-32b-dflash

Status: Open.

#32 In z-lab/dflash;

· gq112 opened on Feb 27, 2026

Plan for large-size Qwen3 or Qwen3.5 models?

Status: Open.

#31 In z-lab/dflash;

· feichendingxiangyong opened on Feb 26, 2026

Performance degradation - Qwen3-Coder-30B-A3B-DFlash

Status: Open.

#30 In z-lab/dflash;

· shahizat opened on Feb 20, 2026

🚀 We just released gpt-oss-20b-DFlash!

Status: Open.

#29 In z-lab/dflash;

· jianc99 opened on Feb 18, 2026

Why fused KV injection instead of independent per-layer injection?

Status: Open.

#26 In z-lab/dflash;

· shreyas269 opened on Feb 9, 2026

[Model Request] Plan for stepfun-ai/Step-3.5-Flash

Status: Open.

#25 In z-lab/dflash;

· mingMelody opened on Feb 9, 2026

Accuracy drops in dflash

Status: Open.

#23 In z-lab/dflash;

· shahizat opened on Feb 8, 2026

Very slow on my nvidia 3090: 44 tok/s

Status: Open.

#20 In z-lab/dflash;

· mp3pintyo opened on Feb 7, 2026

Will the training code be released later?

Status: Open.

#19 In z-lab/dflash;

· jiahe7ay opened on Feb 7, 2026

Training cost

Status: Open.

#18 In z-lab/dflash;

· shx2005 opened on Feb 7, 2026

Qwen 8B from scratch: context length setup + KL vs CE

Status: Open.

#17 In z-lab/dflash;

· Walid-Ahmed opened on Feb 4, 2026