Downloads · 30 days
2.7K
33% of all-time downloads
AIconjured/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-MTP-GGUF-NVFP4
Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-MTP-GGUF-NVFP4 is a text generation model from AIconjured. Use it when you need the model to write or continue text. It is set up for gguf. The card lists the license as other.
Uncensored Qwen3.5 9B (HauhauCS Aggressive) · NVFP4 quantized · MTP head grafted · vision-capable
Downloads · 30 days
2.7K
33% of all-time downloads
All-time downloads
8.3K
Public
Repo size
7.3 GB
Likes
6
Public
Click a slice to open those files.
.gguf7.3 GB · 100%
From the Hugging Face model README
Uncensored Qwen3.5 9B (HauhauCS Aggressive) · NVFP4 quantized · MTP head grafted · vision-capable
A hand-built GGUF of HauhauCS's uncensored Qwen3.5-9B fine-tune, quantized with a custom NVFP4 recipe for Blackwell GPUs, with the Multi-Token-Prediction (MTP) head grafted in from the upstream base model, plus the original vision projector (mmproj) for image input.
This model is built on the work of two teams:
This repository (AIconjured) is the NVFP4 quantization, MTP graft, and packaging of that model.
| File | Size | Description |
|---|---|---|
Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-MTP-GGUF-NVFP4.gguf | 5.9 GB | LLM: 442 tensors (427 main + 15 MTP), block_count=33, nextn_predict_layers=1 |
mmproj-Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-BF16.gguf | 0.9 GB | Vision projector, clip arch, 334 tensors, BF16 (unchanged from upstream) |
Size: ~5.9 GB (LLM) + ~0.9 GB (mmproj) — down from 18.8 GB in BF16.
qwen35 — 32 trunk layers (hybrid DeltaNet SSM + full attention, interval 4), 9.2B params, 262K native context, 4096-dim embeddings, 16 heads / 4 KV headsQwen3.5-9B-MTP-GGUF (Q8_0 carrier)The quantization is a mixed-precision recipe, not a single uniform type. It was built with llama.cpp's --tensor-type-file (regex → type) plus an imatrix generated from ~144 KB of mixed calibration text (PPL ≈ 5.18):
| Type | Tensors | What | Why |
|---|---|---|---|
| NVFP4 | 192 | ffn_gate/up/down, attn_qkv, attn_gate, ssm_out, full-attn attn_q/k/v/output | All the big GEMMs — where the bits are. NVFP4 (FP4 weights + FP8/UE4M3 scales) is the native Blackwell path (BLACKWELL_NATIVE_FP4), fastest on RTX 50-series. |
| Q8_0 | 66 | token_embd, output (LM head), attn_k, ssm_alpha, ssm_beta, ssm_conv1d | Small but accuracy-sensitive tensors; Q8_0 is near-lossless. |
| F32 | 184 | All norm weights, ssm_a, ssm_dt | Norms/decay constants are tiny but quantizing them causes noticeable quality loss; kept at full precision. |
Recipe (llama.cpp --tensor-type-file format):
token_embd\.weight=q8_0
^output\.weight=q8_0
blk\.\d+\.attn_k\.weight=q8_0
blk\.\d+\.ssm_beta\.weight=q8_0
blk\.\d+\.ssm_alpha\.weight=q8_0
blk\.\d+\.ssm_conv1d\.weight=q8_0
blk\.\d+\.attn_output\.weight=nvfp4
blk\.\d+\.ffn_gate\.weight=nvfp4
blk\.\d+\.ffn_up\.weight=nvfp4
blk\.\d+\.ffn_down\.weight=nvfp4
blk\.\d+\.attn_qkv\.weight=nvfp4
blk\.\d+\.attn_gate\.weight=nvfp4
blk\.\d+\.ssm_out\.weight=nvfp4
blk\.\d+\.attn_q\.weight=nvfp4
blk\.\d+\.attn_v\.weight=nvfp4
The HauhauCS fine-tune ships without MTP tensors. The MTP head (15 tensors, blk.32.nextn.*) was grafted from unsloth's Q8_0 base-model carrier, and the metadata was set to block_count=33 + nextn_predict_layers=1 so the loader sees 32 trunk layers + 1 MTP layer at index 32.
Important: the MTP head comes from the base model, not the fine-tune. Speculative decoding is self-verifying (the main model accepts/rejects every draft token), so quality is guaranteed by the uncensored fine-tune — but the accept rate will be lower than a fine-tune-native MTP head, since the base head doesn't perfectly match the fine-tuned distribution.
To actually get the speedup you need an engine that implements MTP spec decode:
llama-server -m Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-MTP-GGUF-NVFP4.gguf \
--mmproj mmproj-Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-BF16.gguf \
--spec-type draft-mtp -c 8192 -ngl 99
Ollama note: Ollama (tested 0.33.3) loads the MTP tensors but does not implement MTP speculative decoding — the tensors are ignored and the model runs as a plain 32-layer model. Text and vision both work fine; you just don't get the draft speedup.
Vision is a separate clip-arch mmproj GGUF (334 tensors, BF16, unmodified from upstream). The LLM file intentionally contains no vision tensors — the qwen35 arch loader rejects them, and this is the structure Ollama expects. Load both files:
FROM ./Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-MTP-GGUF-NVFP4.gguf
FROM ./mmproj-Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-BF16.gguf
The projector uses the qwen3vl_merger with a vision encoder; verified working end-to-end in Ollama (text + image).
378aa2ebc (2026-08-27), CUDA buildllama-quantize --imatrix imatrix-9b.gguf --tensor-type-file recipe.txt ...blk.32.nextn.* tensors + qwen35.nextn_predict_layers=1, qwen35.block_count=33This quantization is a derivative of HauhauCS's uncensored fine-tune of the Qwen3.5-9B base model. See the base model's license at HauhauCS/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive and the Qwen base model's license at QwenLM.