Downloads · 30 days
89
17% of all-time downloads
GestaltLabs/Ornstein-9Bv2.5-NSC-ACE-GGUF
Ornstein-9Bv2.5-NSC-ACE-GGUF is a text generation model from GestaltLabs. Use it when you need the model to write or continue text. It is set up for gguf. The card lists the license as apache-2.0.
llama.cpp GGUF builds of DJLougen/Ornstein-9Bv2.5-NSC-ACE (text decoder). The MTP draft head is embedded in every file (blk.32.nextn.) — enable speculative decoding with --spec-type draft-mtp (llama.cpp b9940+), no se…
Downloads · 30 days
89
17% of all-time downloads
All-time downloads
513
Public
Repo size
48.3 GB
Likes
0
Public
Click a slice to open those files.
.gguf48.2 GB · 100%
From the Hugging Face model README
llama.cpp GGUF builds of DJLougen/Ornstein-9Bv2.5-NSC-ACE (text decoder). The MTP draft head is embedded in every file (blk.32.nextn.*) — enable speculative decoding with --spec-type draft-mtp (llama.cpp b9940+), no separate draft model needed.
| File | Quant | Size | Notes |
|---|---|---|---|
Ornstein-9Bv2.5-NSC-ACE-Q4_K_M.gguf | Q4_K_M | 5.8 GB | smallest, good default |
Ornstein-9Bv2.5-NSC-ACE-Q5_K_M.gguf | Q5_K_M | 6.6 GB | balanced |
Ornstein-9Bv2.5-NSC-ACE-Q6_K.gguf | Q6_K | 7.6 GB | high quality |
Ornstein-9Bv2.5-NSC-ACE-Q8_0.gguf | Q8_0 | 9.8 GB | near-lossless, behaviorally verified |
Ornstein-9Bv2.5-NSC-ACE-F16.gguf | F16 | 18.4 GB | full precision / re-quant source |
llama-server -hf DJLougen/Ornstein-9Bv2.5-NSC-ACE-GGUF:Q8_0 -ngl 99 --spec-type draft-mtp
Measured on an RTX 3090 at Q8_0 with --spec-type draft-mtp: 67.6 → 100.7 tok/s on code (+49%), 90.6 → 110.9 tok/s on prose (+22%) vs. non-speculative on the same build.
Sample with temperature=0.6, top_p=0.95, top_k=20 (Qwen3.5 reasoning defaults). Hermes-style <tools> tool calling works via the embedded chat template. See the main repo for recipe, evaluation, and caveats.