library_name: transformers license: other license_name: glm-5.3 license_link: https://huggingface.co/kitaniai/GLM-5.3-UNCENSORED-MXFP4/blob/main/LICENSE pipeline_tag: text-generation base_model:

• dealignai/GLM-5.3-UNCENSORED-FP8 base_model_relation: quantized tags:
• glm
• glm_moe_dsa
• mxfp4
• quark
• quantized
• mixture-of-experts
• safetensors
• text-generation

Use GLM Uncensored at kitani.ai — hosted inference for your apps and coding tools.

GLM-5.3-UNCENSORED-MXFP4

An MXFP4 quantized version of dealignai/GLM-5.3-UNCENSORED-FP8, published by Kitani.

This release uses AMD Quark's GLM MoE DSA recipe and standard Hugging Face safetensors. Expert matrices use OCP MXFP4; attention, routers, the initial dense MLPs, embeddings, normalization tensors, and the output head remain at higher precision. This is a mixed-precision checkpoint, not a claim that every tensor is four-bit.

Source and provenance

• Immediate source: dealignai/GLM-5.3-UNCENSORED-FP8.
• Exact source revision: 12170112d07e50d7dbe8e5ca1691f4fd8c1179e1.
• The source model card identifies JANGQ-AI/GLM-5.3-FP8 as its base and zai-org/GLM-5.3 as the upstream model.
• This is FP8 → MXFP4 requantization. Source FP8 weights are dequantized using their original block scales before MXFP4 conversion. It does not recover precision already lost in the FP8 source.
• No additional training was performed. The tokenizer and chat template are inherited from the source.
• Conversion completed: 2026-10-06T14:45:24.988739+00:00.

Measured checkpoint statistics

Sizes above count safetensors files, including their headers. Tensor-element counts include non-scale buffers and are not a separately audited trainable-parameter count. Overall storage includes higher-precision tensors and quantization metadata. Exact values are in quantization_stats.json; per-shard sizes and SHA-256 hashes are in artifact_manifest.json.

Quantization details

• Toolkit: AMD Quark 0.13, PyTorch 2.14.1+rocm7.2, ROCm 7.2.53211.
• Method: file-to-file round-to-nearest quantization; no calibration dataset, GPTQ, AWQ, or fine-tuning.
• Weight format: FP4 E2M1 values, two values per byte, with one E8M0 scale per group of 32 consecutive weights.
• Activation policy for quantized layers: dynamic MXFP4, as specified in config.json.
• Quark recipe: LLMTemplate.get("glm_moe_dsa").get_config(scheme="mxfp4"), with *eh_proj additionally excluded to preserve the MTP auxiliary projection in BF16.
• Excluded source FP8 linear weights are recovered to BF16. Existing floating-point excluded tensors retain their source dtype.
• Hardware available: 8× AMD Instinct MI355X; this file-to-file FP8 conversion used one GPU (AMD Radeon Graphics), following Quark's supported single-device FP8 recovery path.

Architecture

The configured context limit is inherited metadata, not a measured serving capacity. Actual context and memory requirements depend on the runtime, KV cache, batch size, and hardware.

Validation

• Passed complete shard/index consistency, file-length, tensor-coverage, dtype, packing-shape, and scale-shape checks.
• Passed independent FP4 decoding and finite-value checks on 64 matrices / 4,980,736 weight elements, sampled across the checkpoint.
• Sampled weight reconstruction relative RMSE vs recovered FP8 source: 0.111689; absolute RMSE: 0.00184817. Sampling uses the first 16 rows of evenly spaced matrix names; this is not a random or exhaustive quality estimate.
• End-to-end inference, perplexity, downstream accuracy, tokens/second, and runtime VRAM consumption have not been measured for this export. Upstream benchmark scores are not results for this quantization.

See sample_validation.json for individual matrices and conversion.json for environment and timing details.

Usage

Use a runtime supporting both GLM-5.3 / glm_moe_dsa and AMD Quark MXFP4. The intended serving path is vLLM with ROCm on supported AMD Instinct hardware. A generic Transformers installation or a runtime supporting only GPT-OSS MXFP4 is not sufficient evidence of compatibility.

Download the checkpoint:

hf download kitaniai/GLM-5.3-UNCENSORED-MXFP4 --local-dir ./GLM-5.3-UNCENSORED-MXFP4

Runtime references: vLLM Quark support and AMD Quark file-to-file quantization. The exact end-to-end serving combination remains unverified for this release.

Reproduce

Install ROCm-enabled PyTorch matching the recorded version, then install amd-quark==0.13, transformers==5.16.0, safetensors==0.8.0, and accelerate.

hf download dealignai/GLM-5.3-UNCENSORED-FP8 --revision 12170112d07e50d7dbe8e5ca1691f4fd8c1179e1 --local-dir ./source
python quantize.py --source ./source --output ./output
python validate.py --source ./source --output ./output --hash
python sample_check.py --source ./source --output ./output

License and attribution

The upstream model uses the GLM-5.3 License, preserved verbatim in this repository. The immediate source repository is tagged MIT, but the upstream model includes its own license conditions; this release is not labeled MIT-only. See LICENSE and LICENSE_PROVENANCE.json. Credit to Z.ai for GLM-5.3, JANGQ-AI for the FP8 base, dealignai for the supplied source checkpoint, and AMD for Quark. Kitani performed this MXFP4 conversion and packaging.