shunyang90/Qwen3-Omni-30B-A3B-Instruct-NVFP4

delete old shard model-00006-of-00006.safetensors

delete old shard model-00005-of-00006.safetensors

delete old shard model-00004-of-00006.safetensors

delete old shard model-00003-of-00006.safetensors

delete old shard model-00002-of-00006.safetensors

delete old shard model-00001-of-00006.safetensors

update index after stripping buffers

strip non-persistent buffers from model-00006-of-00007.safetensors

Upload folder using huggingface_hub

update index after stripping extra buffers

strip extra buffers (inv_freq, code_offset) from model-00007-of-00007.safetensors

strip extra buffers (inv_freq, code_offset) from model-00006-of-00006.safetensors

update index after splitting talker experts

split talker fused MoE experts to per-expert format in model-00006-of-00007.safetensors

split talker fused MoE experts to per-expert format in model-00006-of-00006.safetensors

split talker fused MoE experts to per-expert format in model-00005-of-00007.safetensors

split talker fused MoE experts to per-expert format in model-00005-of-00006.safetensors

Upload folder using huggingface_hub

update index after stripping _amax tensors

strip _amax calibration artifacts from model-00001-of-00006.safetensors

fix: use original tokenizer_config.json with all special tokens

fix: add missing group_size to hf_quant_config.json

Upload folder using huggingface_hub

initial commit