avlp12 / GLM-5.3-Flash-Alis-MLX-6bit

huggingface.co
Total runs: 492
24-hour runs: -51
7-day runs: -666
30-day runs: -200
Model's Last Updated: August 31 2026
text-generation

Introduction of GLM-5.3-Flash-Alis-MLX-6bit

Model Details of GLM-5.3-Flash-Alis-MLX-6bit

GLM-5.3-Flash Alis MLX 6-bit — withdrawn / 회수

Do not download or serve these weights. The checkpoint is being rewritten.

이 가중치를 받거나 서빙하지 마세요. 체크포인트를 재작성 중입니다.

English

This repository previously hosted a stock mlx-lm affine 6-bit conversion of zai-org/GLM-5.3-Flash .

That conversion used a broken mixed-bit recipe :

  • the MoE router ( mlp.gate ) was quantized to 8-bit — the router must not be quantized
  • KDA GEMMs were skipped — those attention-path GEMMs should stay as 8-bit GEMMs

Decode is stuck around 5.5 tok/s . Do not serve this checkpoint, and do not use a previously downloaded copy.

The weight files have been removed from the Hub. The recipe is being fixed and the checkpoint is being rewritten .

This was a stock affine baseline, not an ALIS or DWQ build.

The same bug affects the 4-bit, 6-bit, and 8-bit repos in this set.

한국어

이 저장소에는 원래 zai-org/GLM-5.3-Flash 의 mlx-lm affine 6-bit 변환본이 있었습니다.

그 변환은 잘못된 혼합 비트 레시피 를 썼습니다:

  • MoE 라우터( mlp.gate )를 8-bit로 양자화했습니다. 라우터는 양자화하면 안 됩니다
  • KDA GEMM을 건너뛰었습니다. 어텐션 경로 GEMM은 8-bit로 두어야 합니다

디코드가 약 5.5 tok/s 에 고착됩니다. 서빙에 쓰지 마세요. 이미 받아 둔 복사본도 쓰지 마세요.

가중치 파일은 허브에서 제거했습니다. 레시피를 고친 뒤 체크포인트를 재작성 중 입니다.

ALIS/DWQ 빌드가 아닙니다. 같은 버그가 이 세트의 4-bit / 6-bit / 8-bit 저장소에 모두 있습니다.

Status
Weight shards removed
Checkpoint being rewritten
Source zai-org/GLM-5.3-Flash

Runs of avlp12 GLM-5.3-Flash-Alis-MLX-6bit on huggingface.co

492
Total runs
-51
24-hour runs
-541
3-day runs
-666
7-day runs
-200
30-day runs

More Information About GLM-5.3-Flash-Alis-MLX-6bit huggingface.co Model

More GLM-5.3-Flash-Alis-MLX-6bit license Visit here:

https://choosealicense.com/licenses/mit

GLM-5.3-Flash-Alis-MLX-6bit huggingface.co

GLM-5.3-Flash-Alis-MLX-6bit huggingface.co is an AI model on huggingface.co that provides GLM-5.3-Flash-Alis-MLX-6bit's model effect (), which can be used instantly with this avlp12 GLM-5.3-Flash-Alis-MLX-6bit model. huggingface.co supports a free trial of the GLM-5.3-Flash-Alis-MLX-6bit model, and also provides paid use of the GLM-5.3-Flash-Alis-MLX-6bit. Support call GLM-5.3-Flash-Alis-MLX-6bit model through api, including Node.js, Python, http.

GLM-5.3-Flash-Alis-MLX-6bit huggingface.co Url

https://huggingface.co/avlp12/GLM-5.3-Flash-Alis-MLX-6bit

avlp12 GLM-5.3-Flash-Alis-MLX-6bit online free

GLM-5.3-Flash-Alis-MLX-6bit huggingface.co is an online trial and call api platform, which integrates GLM-5.3-Flash-Alis-MLX-6bit's modeling effects, including api services, and provides a free online trial of GLM-5.3-Flash-Alis-MLX-6bit, you can try GLM-5.3-Flash-Alis-MLX-6bit online for free by clicking the link below.

avlp12 GLM-5.3-Flash-Alis-MLX-6bit online free url in huggingface.co:

https://huggingface.co/avlp12/GLM-5.3-Flash-Alis-MLX-6bit

GLM-5.3-Flash-Alis-MLX-6bit install

GLM-5.3-Flash-Alis-MLX-6bit is an open source model from GitHub that offers a free installation service, and any user can find GLM-5.3-Flash-Alis-MLX-6bit on GitHub to install. At the same time, huggingface.co provides the effect of GLM-5.3-Flash-Alis-MLX-6bit install, users can directly use GLM-5.3-Flash-Alis-MLX-6bit installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

GLM-5.3-Flash-Alis-MLX-6bit install url in huggingface.co:

https://huggingface.co/avlp12/GLM-5.3-Flash-Alis-MLX-6bit

Url of GLM-5.3-Flash-Alis-MLX-6bit

GLM-5.3-Flash-Alis-MLX-6bit huggingface.co Url

Provider of GLM-5.3-Flash-Alis-MLX-6bit huggingface.co

avlp12
ORGANIZATIONS

Other API from avlp12