Native packed NVFP4 conversion of orcarouter/GLM-5.3-Flash-Uncensored-FP8 at revision
55c78761fb78d0f765e2bd01687900d2fd91b5d0. This checkpoint preserves that specific OrcaRouter
abliteration; it is not a quantization of the unmodified Z.ai base model.
lm_head, and the complete layer-45 MTP block.The routed experts use per-16 E2M1 weights, E4M3 block scales, and an FP32
per-tensor scale in standard nvfp4-pack-quantized compressed-tensors format.
For every tensor, the converter compares classic NVFP4 scaling with NVIDIA's
Four-Over-Six M=6/M=4 block-scale search and stores the lower-error result.
The exact OrcaRouter fine-tune is published only as a block-FP8 checkpoint. Its weights are dequantized from their own 128x128 FP8 scale planes before the NVFP4 pass. This preserves the linked fine-tune, but it is necessarily a second lossy quantization step relative to an unavailable BF16 version of that exact fine-tune. Non-expert FP8 weights are materialized as BF16 in this checkpoint.
See conversion-manifest.json and MANIFEST.sha256 for pinned provenance,
tool versions, tensor counts, aggregate error metrics, and file hashes.
model-00002-of-00062.safetensors 10 GBmodel-00001-of-00062.safetensors 8.1 GBmodel-00017-of-00062.safetensors 3.6 GBmodel-00035-of-00062.safetensors 3.2 GBmodel-00028-of-00062.safetensors 3.2 GBmodel-00022-of-00062.safetensors 3.2 GBmodel-00053-of-00062.safetensors 3.2 GBmodel-00005-of-00062.safetensors 3.2 GBmodel-00040-of-00062.safetensors 3.2 GBmodel-00011-of-00062.safetensors 3.2 GBmodel-00046-of-00062.safetensors 3.2 GBmodel-00032-of-00062.safetensors 3.2 GBmodel-00059-of-00062.safetensors 3.2 GBmodel-00024-of-00062.safetensors 3.2 GBmodel-00038-of-00062.safetensors 3.2 GBmodel-00045-of-00062.safetensors 3.2 GBmodel-00052-of-00062.safetensors 3.2 GBmodel-00007-of-00062.safetensors 3.2 GBmodel-00014-of-00062.safetensors 3.2 GBmodel-00021-of-00062.safetensors 3.2 GBmodel-00042-of-00062.safetensors 3.2 GBmodel-00049-of-00062.safetensors 3.2 GBmodel-00004-of-00062.safetensors 3.2 GBmodel-00018-of-00062.safetensors 3.2 GBmodel-00025-of-00062.safetensors 3.2 GBmodel-00039-of-00062.safetensors 3.2 GBmodel-00008-of-00062.safetensors 3.2 GBmodel-00015-of-00062.safetensors 3.2 GBmodel-00036-of-00062.safetensors 3.2 GBmodel-00029-of-00062.safetensors 3.2 GBmodel-00031-of-00062.safetensors 3.2 GBmodel-00043-of-00062.safetensors 3.2 GBmodel-00050-of-00062.safetensors 3.2 GBmodel-00012-of-00062.safetensors 3.2 GBmodel-00019-of-00062.safetensors 3.2 GBmodel-00033-of-00062.safetensors 3.2 GBmodel-00026-of-00062.safetensors 3.2 GBmodel-00054-of-00062.safetensors 3.2 GBmodel-00056-of-00062.safetensors 3.2 GBmodel-00060-of-00062.safetensors 3.2 GBmodel-00057-of-00062.safetensors 3.2 GBmodel-00047-of-00062.safetensors 3.2 GBmodel-00010-of-00062.safetensors 3.2 GBmodel-00003-of-00062.safetensors 3.1 GBmodel-00061-of-00062.safetensors 3.1 GBmodel-00009-of-00062.safetensors 3.0 GBmodel-00048-of-00062.safetensors 3.0 GBmodel-00027-of-00062.safetensors 3.0 GBmodel-00034-of-00062.safetensors 3.0 GBmodel-00041-of-00062.safetensors 3.0 GBmodel-00020-of-00062.safetensors 3.0 GBmodel-00013-of-00062.safetensors 3.0 GBmodel-00006-of-00062.safetensors 3.0 GBmodel-00051-of-00062.safetensors 3.0 GBmodel-00044-of-00062.safetensors 3.0 GBmodel-00030-of-00062.safetensors 3.0 GBmodel-00037-of-00062.safetensors 3.0 GBmodel-00023-of-00062.safetensors 3.0 GBmodel-00016-of-00062.safetensors 3.0 GBmodel-00055-of-00062.safetensors 3.0 GBmodel-00058-of-00062.safetensors 3.0 GBmodel-00062-of-00062.safetensors 1.3 GB有文件?提交种子。
提交种子