← back to catalog · registered 2026-09-15 17:56

ababaka/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU-W4A16-AutoRound

ababaka 27B second-order
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
Downloads · 30-day
209
Likes
0
Model age
today
created 2026-09-15
Downloads over time
Now0from0↑0%
00110 on Sep 150 on Sep 16Sep
Sep 15 → Sep 16 · 2 snapshots · spans 1 day

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

License
apache-2.0
Tags
transformers safetensors qwen3_5 image-text-to-text quantized w4a16 auto-round compressed-tensors vllm uncensored text-generation conversational

Related

Total size
15.6 GB
Files
20
Quantizations
1
Registered
2026-09-15 17:56
Last updated on HF
2026-09-15 17:44

Files by quantization

Auxiliary files 20 files 15.6 GB
model-00004-of-00007.safetensors 3.00 GB eed6f0dc download
model-00001-of-00007.safetensors 2.99 GB 100f7323 download
model-00002-of-00007.safetensors 2.98 GB c930e248 download
model-00003-of-00007.safetensors 2.98 GB 575503c7 download
model-00006-of-00007.safetensors 1.20 GB b1962c7d download
model-00007-of-00007.safetensors 1.20 GB c538c4c8 download
model-00005-of-00007.safetensors 667 MB c4f36b62 download
model_extra_tensors.safetensors 615 MB c2bb24c9 download
mtp_draft_vocab_ids.pt 322 KB 8af90286 download
tokenizer.json 19.1 MB 87a7830d download
model.safetensors.index.json 195 KB 1f1321de download
config.json 21.9 KB 6ad91485 download
quantization_config.json 15.8 KB 263f7f84 download
chat_template.jinja 8.74 KB 5a39aa2b download
README.md 6.53 KB 5a837836 download
.gitattributes 1.53 KB 52373fe2 download
tokenizer_config.json 1.17 KB 4d9ac0cf download
processor_config.json 1.16 KB 33818c7f download
preprocessor_config.json 390 B 2ea84a43 download
generation_config.json 214 B 8b9f95da download

README current version from Hugging Face


license: apache-2.0
library_name: transformers
pipeline_tag: text-generation
tags:

  • qwen3_5
  • quantized
  • w4a16
  • auto-round
  • compressed-tensors
  • vllm
  • uncensored
    base_model: DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU

Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU-W4A16-AutoRound

W4A16 pack-quantized (compressed-tensors, vLLM-ready) version of
DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU
— a Qwen3.8-27B (qwen3_5) finetune, Heretic/uncensored.

Built for single-GPU (RTX 3090 24 GB) vLLM serving with long context: the main
transformer body is int4, lm_head / embeddings / MTP draft module are int8, and a
vocab-truncated draft head (40960 tokens) is included for MTP speculative decoding.

Note on use: the base model is uncensored (refusal behavior removed by its
authors). Responsibility for how you use this quantization lies with you.

What this is

Base model DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU (BF16, ~55 GB)
Architecture Qwen3.5 (qwen3_5), 27B, 64 layers (48 linear-attention DeltaNet + 16 full attention), vocab 248320, MTP module, vision tower, untied embeddings
Quantization W4A16 AutoRound (int4, group 128, symmetric) + int8 lm_head / embed_tokens / MTP
Format compressed-tensors pack-quantized (weight_packed int32 + weight_scale + weight_shape)
Size on disk ~16 GB
Tested with vLLM 0.27.1, transformers 5.15.0, CUDA 13, RTX 3090 24 GB

How much context fits (measured on RTX 3090)

Config KV pool Result
gpu_memory_utilization=0.94, max_model_len=140000 (recommended) 152,173 tokens fits with ~9% headroom
gpu_memory_utilization=0.96, max_model_len=160000 167,692 tokens fits, less runtime headroom
gpu_memory_utilization=0.93, max_model_len=160000 ~146k ceiling does not fit (engine refuses to start)

Pool size varies a few percent between boots; 140k @ 0.94 is the safe everyday setting.
FP8 KV cache (--kv-cache-dtype fp8), prefix caching on, MTP speculative decoding with 3 draft tokens.

Run it (example)

vllm serve ababaka/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU-W4A16-AutoRound \
  --served-model-name qwen3.8-27b \
  --host 0.0.0.0 --port 18020 \
  --gpu-memory-utilization 0.94 \
  --max-model-len 140000 \
  --max-num-seqs 8 \
  --language-model-only \
  --kv-cache-dtype fp8 \
  --mamba-ssm-cache-dtype float16 \
  --async-scheduling \
  --max-num-batched-tokens 2048 \
  --speculative-config '{"method":"mtp","num_speculative_tokens":3,"draft_sample_method":"probabilistic"}' \
  --compilation-config '{"max_cudagraph_capture_size":32,"custom_ops":["+rms_norm","+silu_and_mul"]}' \
  --reasoning-parser qwen3 \
  --enable-auto-tool-choice --tool-call-parser qwen3_xml \
  --enable-prefix-caching --mamba-cache-mode align

For 160k context use --gpu-memory-utilization 0.96 --max-model-len 160000.
--language-model-only skips the vision tower (text serving). The chat template of the
base model is used by default; a Sharp custom template
(peculiar-ragdoll/Qwen-Sharp-Chat-Templates, reasoning_effort via
--default-chat-template-kwargs) was used in testing.

Files

  • model-00001..07-of-00007.safetensors — quantized transformer + int8 lm_head/embed_tokens
  • model_extra_tensors.safetensors — int8 MTP module + mtp.draft_lm_head (40960-token draft head)
  • mtp_draft_vocab_ids.pt — token-id map for the truncated draft vocabulary
  • model.safetensors.index.json, config.json (+ quantization_config.json), tokenizer/processor configs
How this quantization was made (recipe, pipeline, format)

Base

Downloaded DavidAU/...-Heretic-Uncensored-NM-DAU in FP16 (13 shards incl.
model-mtp-restored.safetensors, ~55 GB). Verified same skeleton as the reference
Qwen3.8-27B quant: qwen3_5, 64 layers (48 linear + 16 full), vocab 248320,
tie_word_embeddings=false, mtp.fc + 7 MTP linears, model.visual.* tower.

Quantization (auto-round 0.14.2)

Parameter Value
Toolchain auto-round 0.14.2, transformers 5.15.0, torch 2.13+cu130, compressed-tensors
Scheme W4A16, int4, group 128, symmetric (pack-quantized)
Dataset NeelNanda/pile-10k, 128 samples, seqlen 2048, batch 4
Iters 200
seed / trust_remote_code 42 / True
quant_nontext_module False (vision tower untouched)
Export format llm_compressor
VRAM adaptation low_gpu_mem_usage=True (CPU offload for 24 GB cards)

layer_config kept in BF16: lm_head, linear_attn.in_proj_a/b (all 48 linear-attention
layers), visual.* (whole vision tower), mtp (whole MTP module).
Full run on RTX 3090: 64/64 layers in ~2h41m (~141–159 s/layer), peak VRAM 15.5 GB /
RAM 27.3 GB. Result: 400/607 submodules quantized; ignore list 303 entries
(110 visual + 192 linear_attn + lm_head), no mtp/embed entries — MTP was copied
unquantized into model_extra_tensors.safetensors by the exporter.

Post-quantization pipeline (in this order)

  1. quant_lm_headlm_head → int8 g128 (round-trip rel. error 0.64%, ~1.3 GB VRAM saved)
  2. quant_embedembed_tokens → int8 g128, scales in bf16 (untied embeddings, error 0.60%, ~1.3 GB saved)
  3. quant_mtpmtp.fc + 7 MTP linears → int8 g128 (errors 0.6–1.5%)
  4. build_draft_vocab — draft head sliced from the int8 lm_head to the 40960 most
    frequent token ids → mtp.draft_lm_head.* (213 MB) in model_extra_tensors.safetensors
    • mtp_draft_vocab_ids.pt

Final config_groups (order matters for compressed-tensors matching in vLLM —
first match wins, so the lm_head regex must precede the mtp one):

group_0  targets=["Linear"]             int4 g128 sym   — main body
group_1  targets=["re:.*lm_head$"]      int8 g128       — lm_head + mtp.draft_lm_head
group_2  targets=["re:.*embed_tokens$"] int8 g128       — embed_tokens
group_3  targets=["re:^mtp\\..*"]       int8 g128       — MTP module

ignore: 302 entries (visual + linear_attn norms/projections); lm_head/mtp/embed
are NOT ignored. Verified by loading in vLLM: MTP drafter detected, 40960-token draft
head recognized, coherent generation (no random-token garbage).

License

Base model is Apache 2.0; this quantization follows the base model's license.

README history 2 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-09-15Upload README.md with huggingface_hubb7efe8b6.5 KB
    Loading...
  2. 2026-09-15Upload folder using huggingface_hub46a1ac16.3 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in app" button that hands off directly to a local runtime of your choice - Infrahuman, LM Studio, or Ollama. No API keys, no subscription, no prompt leakage.