nerkyor/Qwen3.6-27B-DSV4Pro-Thinking-Distill
docs: link Lynn Agent releases
split bilingual blobs into monolingual EN/ZH sections (two-paths + GB-vs-GiB note)
GB units (GiB->GB), fix NVFP4 link to HF, add GB-vs-GiB explainer
docs: link Lynn v0.85.6 release
add English NVFP4 quant section (W4A16/W4A4/W4A8 table) to match ZH
add IQ4_XS-imatrix MTP GGUF (14.3GiB, fits 16G no-offload; smoke-verified converging + 31t/s single-stream MTP accept0.80)
Update file-list NVFP4 entry: add W4A4, drop GGUF-for-single-stream
Refine Claude Code section wording
Add Claude Code (experimental) section
Remove modelscope cache artifacts (.ms*/.msc/.mv)
Remove modelscope cache artifacts (.ms*/.msc/.mv)
Remove modelscope cache artifacts (.ms*/.msc/.mv)
Add W4A16+MTP package to main nvfp4/ (w4a4 in standalone NVFP4 repo)
Fix W4A8 size ~13G->~19G
Add nvfp4/ pointer
Add NVFP4 section (3-tier W4A16/W4A4/W4A8 + MTP, device-tagged @R6000)
docs: link Lynn Agent release
card: fix zh-half Q5_K_M base-GPQA streaming caveat
Add NVFP4 W4A16-style variant link to Files section
add Q4_K_S-imatrix MTP GGUF
add Q5_K_M eval section (distill vs base, streaming harness)
add F16/BF16 GGUF (MTP) as ROCm FPX quant source
Card: note BF16/FP8 safetensors also support MTP via vLLM/SGLang
Add MTP (nextn) head — BF16 now supports vLLM/SGLang speculative
Add MTP (nextn) head — BF16 now supports vLLM/SGLang speculative
Add MTP (nextn) head — BF16 now supports vLLM/SGLang speculative
Add vision mmproj (F16) for multimodal in llama.cpp
Make vLLM-loadable: official multimodal config + vision + preprocessor
Make vLLM-loadable: official multimodal config + vision + preprocessor
Make vLLM-loadable: official multimodal config + vision + preprocessor
Make vLLM-loadable: official multimodal config + vision + preprocessor
Make vLLM-loadable: official multimodal config + vision + preprocessor
Simplify bilingual header
add fp8/vocab.json
add fp8/video_preprocessor_config.json
add fp8/tokenizer_config.json
add fp8/tokenizer.json
add fp8/preprocessor_config.json
add fp8/mtp.safetensors
add fp8/model.safetensors.index.json
add fp8/model-vision.safetensors
add fp8/model-00012-of-trunk.safetensors
add fp8/model-00011-of-trunk.safetensors
add fp8/model-00010-of-trunk.safetensors
add fp8/model-00009-of-trunk.safetensors
add fp8/model-00008-of-trunk.safetensors
add fp8/model-00007-of-trunk.safetensors
add fp8/model-00006-of-trunk.safetensors
add fp8/model-00005-of-trunk.safetensors
add fp8/model-00004-of-trunk.safetensors