Deviad
/

DeepSeek-V4-Flash-MLX-Q4Q8

Text Generation

mixture-of-experts

Mixture of Experts

4-bit precision

8-bit precision

Model card Files Files and versions

Deviad commited on 8 days ago

Commit

1fcb87f

·

verified ·

1 Parent(s): 78a20cd

Upload README.md with huggingface_hub

Files changed (1) hide show

README.md +2 -2

README.md CHANGED Viewed

@@ -237,7 +237,7 @@ gibberish. Do not use that script.
 The full pipeline (download → convert → re-quantize → finalize → patch
 → verify) is automated in
-[`build_mlx_q4q8.sh`](https://github.com/...) (companion script in the
 project repo). Quick reference of the steps:
 ```
@@ -256,7 +256,7 @@ project repo). Quick reference of the steps:
 M3 Ultra: ~75 minutes plus the initial download (~160 GB at ~150 MB/s =
 ~18 minutes on a fast link).
-See [`requantization-plan.md`](https://github.com/...) for the
 diagnostic write-up of why the requantize step is needed.
 ## License & attribution

 The full pipeline (download → convert → re-quantize → finalize → patch
 → verify) is automated in
+[`build_mlx_q4q8.sh`](https://huggingface.co/Deviad/DeepSeek-V4-Flash-MLX-Q4Q8/blob/main/build_mlx_q4q8.sh) (companion script in the
 project repo). Quick reference of the steps:
 ```
 M3 Ultra: ~75 minutes plus the initial download (~160 GB at ~150 MB/s =
 ~18 minutes on a fast link).
+See [`requantization-plan.md`](https://huggingface.co/Deviad/DeepSeek-V4-Flash-MLX-Q4Q8/blob/main/requantization-plan.md) for the
 diagnostic write-up of why the requantize step is needed.
 ## License & attribution