singulared/Ornith-1.0-35B-MTP-GGUF

Ornith-1.0-35B-MTP (GGUF, Q8_0)

Ornith-1.0-35B (DeepReinforce) with an embedded MTP (Multi-Token Prediction / nextn) head grafted in, enabling self-speculative decoding in llama.cpp for ~+30% decode throughput at identical output quality.

Why

Ornith-1.0-35B is an agentic coder fine-tuned from Qwen3.6-35B-A3B (same qwen35moe architecture, same tokenizer). The base Qwen ships with an embedded MTP head; Ornith's release does not — so it decodes without self-speculation (~50 t/s). Because the fine-tune barely shifts the relevant hidden states, the base model's MTP head transfers almost perfectly when grafted in.

Measured (llama.cpp Vulkan, Radeon 8060S / Strix Halo)

	decode t/s	draft acceptance
Ornith-1.0-35B Q8_0 (no MTP)	~50	—
this model (Q8_0 + grafted MTP)	63–66	0.859 (mean accepted draft len 3.05 / 3)

The weights are unchanged → output quality is identical; the speedup is pure self-speculation.

Usage (llama.cpp)

llama-server -m ornith-1.0-35b-MTP-Q8_0.gguf \
  --spec-type draft-mtp --spec-draft-n-max 3 --spec-draft-p-min 0.6 \
  -fa on -ngl 99 -c 131072 --jinja --alias ornith

The MTP head is embedded — no separate draft model (-md) required.

How it was made (reproducible)

The donor MTP block (blk.40 = a full nextn layer: attention + MoE experts + nextn.{eh_proj, enorm, hnorm, shared_head_norm}) was spliced from Qwen3.6-35B-A3B-Q8_0.gguf into Ornith-1.0-35B-Q8_0.gguf with gguf-py:

1. Copy all of Ornith's tensors + metadata (raw quantized round-trip — no dequant/requant).
2. Append Qwen's 20 blk.40.* tensors.
3. Set qwen35moe.block_count = 41 and add qwen35moe.nextn_predict_layers = 1.

Both bases share the architecture + tokenizer, so the head plugs in directly with no retraining.

Licensing & attribution

A derivative of two permissively-licensed models; both are credited and their licenses apply to their respective parts:

• Ornith-1.0-35B — © DeepReinforce — MIT — the base weights (733 of 753 tensors).
• Qwen3.6-35B-A3B — © Alibaba Cloud / Qwen — Apache-2.0 — the grafted MTP (blk.40) head.

Quantization: Q8_0. Not affiliated with or endorsed by DeepReinforce or the Qwen team.