Abliterated Vontra GLM-5.3-Flash MLX oQ4 for Mac Studio oMLX 0.6.3rc2 Dual-ANE. Same Dealign o_proj L15–45 transplant as the Spark NVFP4 pack — L0–14 stay stock, MTP included.
See RESPONSIBLE_USE.md and the gate form above. Access is gated with automatic approval after you agree.
| HF (these weights) | https://huggingface.co/drowzeys/keys-Mac-oMLX-0.6.3.2RC-Dual-ANE-GLM-5.3-Flash-Abliterated-oQ4 |
| GitHub (oMLX Dual-ANE recipe) | https://github.com/drowzeys/keys-Mac-oMLX-0.6.3.2RC-Dual-ANE-GLM-5.3-Flash-Abliterated-oQ4 |
| Spark NVFP4 cousin (same ablit) | HF · GitHub |
| Stock oQ4 parent | Vontra/GLM-5.3-Flash-MLX-oQ4-MTP |
Ablit source (o_proj L15–45) |
dealignai/GLM-5.3-Flash-UNCENSORED-NVFP4 |
| Upstream | zai-org/GLM-5.3-Flash |
| Ablit | L15–45 self_attn.o_proj as BF16 (31 tensors, includes MTP) · L0–14 affine oQ4 stock |
| Gate | 32/32 bypass, 0 refuse, 0 garble |
| Decode (M3 Ultra 256 GB, oMLX 0.6.3rc2) | ~24 tok/s (192-token gens 22.7–25.0; 128-token 23.4 / 23.9). Stock oQ4 on the same box: 26.4 tok/s. |
These weights have safety refusals removed. Research / red-team only — you supply the guardrails.
Blackfrost-style rank-1 projection + affine requant stayed at 19/32 on this quant. The Spark 32/32 dest is a byte-copy of Dealign o_proj. On MLX we do not requantize those tensors: they land as BF16 nn.Linear weights (model-ablit-oproj-l15-45.safetensors). Experts, vision, QKV, and L0–14 o_proj stay Vontra affine oQ4.
git clone https://github.com/drowzeys/keys-Mac-oMLX-0.6.3.2RC-Dual-ANE-GLM-5.3-Flash-Abliterated-oQ4.git
cd keys-Mac-oMLX-0.6.3.2RC-Dual-ANE-GLM-5.3-Flash-Abliterated-oQ4
# Hugging Face gate: agree on the model card, then `hf auth login`
bash oneshot-setup.sh
omlx serve --model-dir ~/.omlx/models --host 0.0.0.0 --port 11500
Serve id: glm53-flash-oq4-mtp-ablit-l15-45. Dual-ANE tile 4096, Lightning MTP on, thinking off.
Cite the original authors first.
| Who | What we used |
|---|---|
| Z.ai | Upstream GLM-5.3-Flash |
| Vontra | Mac oQ4 body; L0–14 o_proj, experts, vision stay theirs |
| dealignai (compute: Jordan Schenck) | BF16 o_proj L15–45 + MTP that reaches 32/32 |
| jundot | oMLX, oQ, Lightning MTP, Dual-ANE |
| onthehub97 | Dual-ANE/GPU prompt processing |
| Blaizzy | day-0 glm5_next in mlx-vlm |
| PipeNetwork | ClampedSwiGLU / Mac glm5_next runtime |
| ml-explore | MLX |
| Blackfrost | Direction reference; not this checkpoint |
| LibertAI | Spark NVFP4 layout Dealign o_proj was extracted from |
Pack: drowzeys / keys. Donate: GoFundMe.
model-00003-of-00035.safetensors 5.7 GBmodel-00009-of-00035.safetensors 5.4 GBmodel-00011-of-00035.safetensors 5.4 GBmodel-00015-of-00035.safetensors 5.4 GBmodel-00018-of-00035.safetensors 5.4 GBmodel-00020-of-00035.safetensors 5.4 GBmodel-00027-of-00035.safetensors 5.4 GBmodel-00010-of-00035.safetensors 5.4 GBmodel-00012-of-00035.safetensors 5.4 GBmodel-00017-of-00035.safetensors 5.4 GBmodel-00024-of-00035.safetensors 5.4 GBmodel-00028-of-00035.safetensors 5.4 GBmodel-00030-of-00035.safetensors 5.4 GBmodel-00008-of-00035.safetensors 5.4 GBmodel-00021-of-00035.safetensors 5.4 GBmodel-00023-of-00035.safetensors 5.4 GBmodel-00026-of-00035.safetensors 5.4 GBmodel-00029-of-00035.safetensors 5.4 GBmodel-00013-of-00035.safetensors 5.4 GBmodel-00014-of-00035.safetensors 5.4 GBmodel-00019-of-00035.safetensors 5.4 GBmodel-00025-of-00035.safetensors 5.4 GBmodel-00032-of-00035.safetensors 5.4 GBmodel-00033-of-00035.safetensors 5.4 GBmodel-00016-of-00035.safetensors 5.4 GBmodel-00005-of-00035.safetensors 5.4 GBmodel-00006-of-00035.safetensors 5.4 GBmodel-00022-of-00035.safetensors 5.4 GBmodel-00031-of-00035.safetensors 5.4 GBmodel-00004-of-00035.safetensors 5.4 GBmodel-00007-of-00035.safetensors 5.4 GBmodel-00034-of-00035.safetensors 5.4 GBmodel-00001-of-00035.safetensors 4.7 GBmodel-00002-of-00035.safetensors 3.9 GBmodel-ablit-oproj-l15-45.safetensors 2.7 GBmodel-00035-of-00035.safetensors 1.6 GBVontra/GLM-5.3-Flash-MLX-oQ4-MTP