Qwen3.5-35B-A3B-Unredacted-MAX

Qwen3.5-35B-A3B-Unredacted-MAX is an unredacted evolution built on top of Qwen/Qwen3.5-35B-A3B. This model applies advanced refusal direction analysis and abliterated training strategies to reduce internal refusal behaviors while preserving the reasoning and instruction-following strengths of the original architecture. The result is a powerful 35B parameter Mixture-of-Experts language model optimized for detailed responses and improved instruction adherence.

This model is materialized for research and learning purposes only. The model has reduced internal refusal behaviors, and any content generated by it is used at the user’s own risk. The authors and hosting page disclaim any liability for content generated by this model. Users are responsible for ensuring that the model is used in a safe, ethical, and lawful manner.

Key Highlights

• Advanced Refusal Direction Analysis: Uses targeted activation analysis to identify and mitigate refusal directions within the model’s latent space.
• Unredacted MAX Training: Fine-tuned to significantly reduce refusal patterns while maintaining coherent and detailed outputs.
• 35B MoE Architecture (A3B): Built on Qwen3.5-35B-A3B, offering stronger reasoning, scalability, and efficiency through Mixture-of-Experts design.
• Improved Instruction Adherence: Optimized to follow complex prompts with minimal unnecessary refusals.
• High-Capability Deployment: Suitable for advanced research experimentation, powerful local inference setups, and large-scale AI applications.

Quick Start with Transformers

pip install transformers==5.3.0 (or) git+https://github.com/huggingface/transformers.git

from transformers import Qwen3_5MoeForConditionalGeneration, AutoProcessor
import torch

model = Qwen3_5MoeForConditionalGeneration.from_pretrained(
    "prithivMLmods/Qwen3.5-35B-A3B-Unredacted-MAX",
    torch_dtype="auto",
    device_map="auto"
)

processor = AutoProcessor.from_pretrained(
    "prithivMLmods/Qwen3.5-35B-A3B-Unredacted-MAX"
)

messages = [
    {
        "role": "user",
        "content": [
            {"type": "text", "text": "Explain how transformer models work in simple terms."}
        ],
    }
]

text = processor.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True
)

inputs = processor(
    text=[text],
    padding=True,
    return_tensors="pt"
).to("cuda")

generated_ids = model.generate(**inputs, max_new_tokens=256)

generated_ids_trimmed = [
    out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]

output_text = processor.batch_decode(
    generated_ids_trimmed,
    skip_special_tokens=True,
    clean_up_tokenization_spaces=False
)

print(output_text)

Intended Use

• Alignment & Refusal Research: Studying refusal behaviors and the impact of activation-level modifications.
• Red-Teaming Experiments: Evaluating robustness across adversarial or edge-case prompts.
• High-Capability Local AI Deployment: Running powerful instruction models on high-memory GPUs or multi-GPU setups.
• Research Prototyping: Experimentation with large transformer architectures and alignment techniques.

Limitations & Risks

Important Note: This model intentionally reduces built-in refusal mechanisms.

• Sensitive Output Possibility: The model may generate controversial or explicit responses depending on prompts.
• User Responsibility: Outputs should be handled responsibly and within legal and ethical boundaries.
• Compute Requirements: A 35B MoE model requires substantial GPU memory or optimized inference techniques (quantization, tensor parallelism).

Dataset & Acknowledgements

• Uncensor any LLM with Abliteration – by Maxime Labonne
• harmful_behaviors and harmless_alpaca – by Maxime Labonne
• Remove Refusals with Transformers (a proof-of-concept implementation to remove refusals from an LLM without using TransformerLens) – by Sumandora
• LLM-LAT/harmful-dataset – by LLM Latent Adversarial Training