---

license: mit

base_model:

• EryriLabs/Ornith-1.5-35B-A3B-BigBang-MTP

tags:

• gguf
• llama.cpp
• moe
• mtp
• speculative-decoding

---

Ornith-1.5-35B-A3B-BigBang-MTP — GGUF

GGUF quants of EryriLabs/Ornith-1.5-35B-A3B-BigBang-MTP:

a TIES merge of Ornith-1.5-35B-A3B + BigBang-v1 over Qwen3.6-35B-A3B, with Ornith's **untrained placeholder MTP head

replaced by Qwen3.6's trained one** — so speculative decoding actually works (~75% draft acceptance vs ~13% for the

stock random head; see the main repo card for the measurements).

Requires a recent llama.cpp (b10507+ tested; the Qwen3.5-MoE hybrid arch is new).

Where is the MTP?

Two options, pick one:

1. Separate draft (recommended): the main GGUFs deliberately contain no MTP tensors. Download mtpdraft-Q8_0.gguf (2 GB) once and pass it with -md — it works with every quant here, and with stock Ornith-1.5 GGUFs too.
2. Embedded: Ornith-1.5-35B-A3B-BigBang-MTP-Q4_K_M-embedded.gguf carries the trained head inside the main file (stock-style blk.40 layout), for tools that expect it embedded.

If you inspect a non-embedded main GGUF and see no MTP tensors, that is by design, not a missing feature.

Files

(BF16 reference weights live in the main safetensors repo.)

| file | size | note |

|---|---|---|

| Ornith-1.5-35B-A3B-BigBang-MTP-Q4_K_M.gguf | 21.2 GB | fits a single 24 GB GPU |

| Ornith-1.5-35B-A3B-BigBang-MTP-Q5_K_M.gguf | 24.7 GB | |

| Ornith-1.5-35B-A3B-BigBang-MTP-Q6_K.gguf | 28.5 GB | recommended if you have the VRAM |

| Ornith-1.5-35B-A3B-BigBang-MTP-Q8_0.gguf | 36.9 GB | |

| mtpdraft-Q8_0.gguf | 2.0 GB | MTP speculative draft — pass with -md |

| Ornith-1.5-35B-A3B-BigBang-MTP-Q4_K_M-embedded.gguf | ~21.7 GB | Q4_K_M with the trained MTP head embedded (stock-style) |

Usage

The MTP draft is optional. The main GGUFs contain no MTP tensors and work as normal models on their own (LM Studio, Ollama, plain llama.cpp):

llama-server -m Ornith-1.5-35B-A3B-BigBang-MTP-Q4_K_M.gguf -ngl 99 -c 4096 --jinja -fa on

Add the draft for ~+29% generation speed (outputs are unchanged; speculative decoding only affects speed):

llama-server -m Ornith-1.5-35B-A3B-BigBang-MTP-Q4_K_M.gguf \
  -md mtpdraft-Q8_0.gguf --spec-draft-n-max 4 --spec-draft-n-min 1 \
  -ngl 99 -c 4096 --jinja -fa on

mtpdraft-Q8_0.gguf also works as a -md draft for stock Ornith-1.5 GGUFs (same tower).

Reasoning is always on (<think>), so allow generous max_tokens. Suggested sampling: temp 0.1, top_k 50.

On a single RTX 3090 (Q4_K_M + Q8_0 draft) the draft lifts generation from 131 to 169 tok/s (+29%), with 55–75% acceptance (highest on code).

Merged and measured by EryriLabs (Dwain Barnes), 2026-08-20.