Hugging Face's logo

Spaces:
halley-ai
/
README
No application file

App Files Community
1
sebastavar commited on
Commit
7cbd15f
·
verified ·
1 Parent(s): 0333bcf

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -2
README.md CHANGED
@@ -18,12 +18,11 @@ High-quality, Apple-Silicon–optimized **MLX** builds, tools, and evals — foc
18
  ## 🚀 Featured models
19
 
20
  | Repo | Bits / GS | Footprint | Notes |
21
- |---|---:|---|---:|---|
22
  | **HalleyAI/gpt-oss-20b-MLX-4bit-gs32** | Q4 / 32 | ~13.1 GB | Best speed on 32 GB; near-baseline quality (+1.81% PPL vs 8-bit) |
23
  | **HalleyAI/gpt-oss-20b-MLX-6bit-gs32** | Q6 / 32 | ~18.4 GB | Near-Q8 fidelity (-0.51% PPL vs 8-bit) |
24
  | **Reference (8-bit)** | Q8 / 32 | — | Use upstream: `lmstudio-community/gpt-oss-20b-MLX-8bit` |
25
 
26
  > **Format:** MLX (not GGUF). For Linux/Windows or non-MLX stacks, use a GGUF build with llama.cpp.
27
 
28
- ---
29
 
 
18
  ## 🚀 Featured models
19
 
20
  | Repo | Bits / GS | Footprint | Notes |
21
+ |---|---:|---:|---|
22
  | **HalleyAI/gpt-oss-20b-MLX-4bit-gs32** | Q4 / 32 | ~13.1 GB | Best speed on 32 GB; near-baseline quality (+1.81% PPL vs 8-bit) |
23
  | **HalleyAI/gpt-oss-20b-MLX-6bit-gs32** | Q6 / 32 | ~18.4 GB | Near-Q8 fidelity (-0.51% PPL vs 8-bit) |
24
  | **Reference (8-bit)** | Q8 / 32 | — | Use upstream: `lmstudio-community/gpt-oss-20b-MLX-8bit` |
25
 
26
  > **Format:** MLX (not GGUF). For Linux/Windows or non-MLX stacks, use a GGUF build with llama.cpp.
27
 
 
28