base_model: Jackrong/Qwopus3.6-27B-Coder
base_model_relation: quantized
license: apache-2.0
library_name: transformers
pipeline_tag: image-text-to-text
language:
- en
- zh
tags: - nvfp4
- fp4
- w4a4
- gptq
- quantized
- compressed-tensors
- llm-compressor
- vllm
- qwen3_5
- vision-language
- thinking
- code
- coder
- uncensored
- abliterated
- heretic
| Params | 27B |
|---|---|
| Active | 27B (dense) |
| Size | 18 GB |
| Perplexity | 6.73 |
| Refusals | 99 → 3 / 100 |
| Context | 256K |
| MTP head | bf16 |
TL;DR: Qwopus3.6-27B-Coder abliterated, quantized to NVFP4 (W4A4) for vLLM on NVIDIA Blackwell. 18 GB, wikitext-2 PPL 6.73, 256K agentic coder, refusals removed (2-round Heretic).
Qwopus3.6-27B-Coder abliterated NVFP4
Jackrong/Qwopus3.6-27B-Coder,
abliterated (refusal direction removed) with Heretic in
two iterative rounds, then quantized to NVFP4 (W4A4) in the compressed-tensorsnvfp4-pack-quantized format with llm-compressor
(GPTQ + MSE, shared fused-layer scales).
Near-lossless and decensored. Two Heretic rounds cut refusals from 99/100 to 3/100 of
held-out harmful prompts while keeping a KL divergence of 0.0398 to the original model
(well under the 0.5 line that signals capability damage). NVFP4 then compresses to ~18 GB with a
wikitext-2 perplexity of 6.73.
- Built for vLLM on NVIDIA Blackwell (4-bit weight + 4-bit activation). Pre-Blackwell GPUs
run it weight-only. - Loading and generation verified in vLLM on an NVIDIA GB10 (Blackwell, sm_121).
Uncensored / abliterated model. It follows instructions without refusal guardrails. The
abliteration only removes refusals; all other behaviour comes from the base model. You
are responsible for how you use it.
Fidelity
Near-lossless versus the bf16 source, 18 GB vs 55.6 GB bf16 (~33%), at wikitext-2 perplexity 6.73 and KL divergence 0.0398 to the original. GPTQ error compensation and an MSE observer keep the drop from bf16 minimal; the header lists the full characteristics and Quantization covers the recipe.
Quickstart
NVFP4 is auto-detected from config.json (compressed-tensors); no quantization flag
needed. --reasoning-parser qwen3 splits the <think> block into reasoning_content;--tool-call-parser qwen3_coder enables tool/function calling for agentic coding.
vllm serve maci0/Qwopus3.6-27B-Coder-abliterated-NVFP4 \
--served-model-name qwopus-27b-coder-abliterated-nvfp4 \
--max-model-len 131072 \
--gpu-memory-utilization 0.90 \
--kv-cache-dtype fp8 \
--reasoning-parser qwen3 \
--enable-auto-tool-choice --tool-call-parser qwen3_coder
- Supports up to 262144 tokens; keep at least 128K to preserve thinking quality.
--max-model-len 131072is a safe default; raise it if memory allows. - Add
--language-model-onlyto skip the vision tower and free KV cache for text use. - The parser flags are not auto-detected; pass them explicitly.
Python (OpenAI client)
from openai import OpenAI
client = OpenAI(base_url="http://localhost:8000/v1", api_key="x")
r = client.chat.completions.create(
model="qwopus-27b-coder-abliterated-nvfp4",
messages=[{"role": "user", "content": "Write a Python function that merges two sorted lists."}],
)
print(r.choices[0].message.content)
curl
curl http://localhost:8000/v1/chat/completions -H "Content-Type: application/json" -d '{
"model": "qwopus-27b-coder-abliterated-nvfp4",
"messages": [{"role": "user", "content": "Write a Python function that merges two sorted lists."}]
}'
About the base model
A 27B Qwen3.5-family vision-language model specialized for code (Qwopus 3.6 Coder),
with thinking-mode reasoning and a 256K context window.
- 64 decoder layers: hybrid gated delta-net linear attention plus full attention, dense
MLP, plus a vision tower for image and video input. - 256K context (
max_position_embeddings262144). - Thinking mode by default, with an instruct toggle.
Abliteration
Heretic removes the refusal direction with a TPE-optimized
search over per-component ablation strength, jointly minimizing refusal rate and KL divergence
from the original model, then merges the best trial. This model was abliterated in two iterative
rounds: round 1 removed the dominant refusal direction, then Heretic was re-run on the round-1
model to remove the residual refusal direction that surfaced once the first was gone. Because Qwopus
is a thinking model, evaluation ran in non-thinking mode so each judged response is a real answer
rather than an unfinished <think> block; the shipped model restores the original thinking chat
template.
| Round | Refusals | Note |
|---|---|---|
| Baseline | 99/100 | original model |
| Round 1 | 25/100 | dominant refusal direction removed |
| Round 2 | 3/100 | residual direction removed (final) |
- Datasets:
mlabonne/harmless_alpaca(good) vsmlabonne/harmful_behaviors(bad). - Final KL divergence 0.0398 (capability preserved, well under the 0.5 damage line).
Quantization
| Scheme | NVFP4, W4A4 |
| Weight rounding | GPTQ (Hessian-based error compensation), MSE observer |
| Weights | FP4 (E2M1), group_size=16, tensor_group, FP8 (E4M3) group scales, shared across fused layers |
| Activations | FP4, dynamic per-group, FP8 (E4M3) scales |
| Quantized | all language-model Linear layers |
| Kept in bf16 | vision tower (model.visual.*), lm_head, MTP head |
| Untouched | gated delta-net Conv1d and SSM params (A_log, dt_bias), never Linear |
GPTQ is a quantization-time cost only; inference speed and format are identical to
plain round-to-nearest NVFP4, but it chooses better 4-bit values.
Calibration: 512 domain-matched samples (long reasoning + general chat + code),max_seq_len=2048, text-only path through the VL model.
Recommended sampling
Thinking mode is the default.
- Thinking, precise coding:
temperature=0.6,top_p=0.95,top_k=20 - Thinking, general:
temperature=1.0,top_p=0.95,top_k=20 - Instruct / non-thinking:
temperature=0.7,top_p=0.80,top_k=20 - To run non-thinking, set
{%- set enable_thinking = false %}in the chat template, or
passextra_body={"chat_template_kwargs": {"enable_thinking": false}}.
Related
- Base model: Jackrong/Qwopus3.6-27B-Coder
- Space: Rogue Quants
- Collection: NVFP4 Quants
- Sibling NVFP4 quants:
Notes
- Needs NVIDIA Blackwell (sm_121, e.g. GB10) for accelerated W4A4; pre-Blackwell GPUs run it weight-only.
--reasoning-parserand--tool-call-parserare not auto-detected; pass them explicitly.- Thinking mode is on by default; toggle it via the chat template or
chat_template_kwargs. - No refusal guardrails; you are responsible for how you use it.
License
Apache-2.0, following the base model. Intended use and all responsibility for use follow
the base model.
Credits
- Base model: Jackrong
- Abliteration: Heretic by Philipp Emanuel Weidmann
- Quantization tooling: llm-compressor / compressed-tensors