Qwen3.5-9B LoRA SFT distillation: R7 (86.8% eval) + R8 calibration. Datasets, FP16 checkpoints, and pipeline docs.
lee
cudabenchmarktest
AI & ML interests
Finetuning small language models, maintaining quality chain of thought, refusal and abliteration, along with novel reasoning distillation techniques. When you are wrestling for possession of a sword, the man with the handle always wins.
Recent Activity
updated a dataset 2 days ago
cudabenchmarktest/r8b-tool-sft updated a dataset 2 days ago
cudabenchmarktest/r8-eval-suite-5bucket updated a dataset 2 days ago
cudabenchmarktest/r8-thinking-fix-sftOrganizations
None yet
models 11
cudabenchmarktest/qwen3.5-9b-r5-research-GGUF
Text Generation • 9B • Updated • 1.17k
cudabenchmarktest/qwen3.5-9b-qwen3.6-reasoning-distilled-GGUF
Text Generation • 10B • Updated • 5.33k • 1
cudabenchmarktest/qwen3.5-9b-qwen3.6-reasoning-distilled
Text Generation • Updated • 266
cudabenchmarktest/qwen3.5-9b-r7-research-vision
Image-Text-to-Text • 9B • Updated • 31
cudabenchmarktest/qwen3.5-9b-r7-research
Text Generation • 9B • Updated • 447
cudabenchmarktest/personaplex-7b-nf4-distilled
Updated
cudabenchmarktest/personaplex-7b-hybrid
Updated
cudabenchmarktest/personaplex-7b-nf4
Updated • 81
cudabenchmarktest/personaplex-7b-turbo2bit
Updated • 143 • 4
cudabenchmarktest/Qwen3.5-27B-turbo2bit
8B • Updated • 19
datasets 6
cudabenchmarktest/r8b-tool-sft
Viewer • Updated • 1.6k • 21
cudabenchmarktest/r8-eval-suite-5bucket
Viewer • Updated • 192 • 44
cudabenchmarktest/r8-thinking-fix-sft
Viewer • Updated • 2.9k • 33
cudabenchmarktest/r7-additive-sft
Viewer • Updated • 4.04k • 27
cudabenchmarktest/r8-calibration-sft
Viewer • Updated • 7.64k • 35
cudabenchmarktest/r9-research-framework
Viewer • Updated • 1.55k • 43