prithivMLmods

VibeThinker-3B-heretic_decensored

Key Highlights

Heretic-Based Abliteration: Modified using the Heretic toolkit to identify and alter refusal-related representations within the model.
Reduced Refusal Behavior: Optimized to minimize internal refusal tendencies while maintaining reasoning performance.
VibeThinker Backbone: Built directly on top of WeiboAI/VibeThinker-3B.
Reasoning-Oriented Performance: Preserves advanced mathematical, coding, and STEM reasoning capabilities after abliteration.
Research-Focused Release: Designed for alignment research, model behavior analysis, and evaluation of refusal-direction modifications.
Efficient 3B Deployment: Suitable for local inference, research environments, and resource-constrained deployment setups.

Model Lineage

Model Path: prithivMLmods/VibeThinker-3B-heretic_decensored
Intermediate Base Model: WeiboAI/VibeThinker-3B by WeiboAI
Foundation Model: Qwen/Qwen2.5-Coder-3B by Qwen

Abliteration Parameters

Table with columns: Parameter, Value
Parameter	Value
direction_index	21.88
attn.o_proj.max_weight	1.37
attn.o_proj.max_weight_position	21.25
attn.o_proj.min_weight	1.36
attn.o_proj.min_weight_distance	19.61
mlp.down_proj.max_weight	1.49
mlp.down_proj.max_weight_position	31.01
mlp.down_proj.min_weight	1.48

Performance

Table with columns: Metric, This model, Original model (WeiboAI/VibeThinker-3B)
Metric	This model	Original model (WeiboAI/VibeThinker-3B)
KL divergence	0.0933	0 (by definition)
Refusals	6/100	64/100

Quick Start with Transformers

bash

pip install transformerspip install accelerate

python

from transformers import AutoTokenizer, AutoModelForCausalLMimport torch model = AutoModelForCausalLM.from_pretrained(    "prithivMLmods/VibeThinker-3B-heretic_decensored",    torch_dtype="auto",    device_map="auto") tokenizer = AutoTokenizer.from_pretrained(    "prithivMLmods/VibeThinker-3B-heretic_decensored") messages = [    {        "role": "user",        "content": "Explain how a transformer model processes text."    }] inputs = tokenizer.apply_chat_template(    messages,    tokenize=True,    add_generation_prompt=True,    return_tensors="pt").to(model.device) outputs = model.generate(    inputs,    max_new_tokens=512) print(    tokenizer.decode(        outputs[0][inputs.shape[-1]:],        skip_special_tokens=True    ))

Intended Use

Alignment Research: Studying refusal-direction analysis and behavior modification techniques.
Model Evaluation: Benchmarking reasoning, instruction-following, and safety-related behaviors.
Red Teaming: Analyzing model responses under reduced-refusal conditions.
Mathematical Reasoning Research: Evaluating performance on verifiable reasoning tasks.
Coding and STEM Evaluation: Studying behavior across programming and scientific reasoning domains.
Local Deployment: Running capable reasoning models on consumer hardware and research environments.

Limitations & Risks

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

Sensitive Content Risk: May generate unrestricted, controversial, or unsafe outputs.
User Responsibility: Requires careful and ethical use.
Experimental Modifications: Behavior may differ significantly from the original model.
Alignment Trade-offs: Reduced refusal behavior may impact safety filtering and response constraints.
Potential Artifacts: Certain prompts may expose unexpected outputs resulting from the abliteration process.
Reasoning Biases: The model may inherit strengths and limitations from the underlying VibeThinker-3B training process.

Acknowledgements

Heretic: Fully automatic censorship removal framework for language models. This project was used to perform the refusal-direction analysis and ablation procedures that form the foundation of this model.
WeiboAI/VibeThinker-3B: The intermediate base model providing the reasoning capabilities used in this release.
Qwen/Qwen2.5-Coder-3B: The foundation model upon which VibeThinker-3B was originally built.
Model Trials & Evaluation: Experimental evaluations, refusal measurements, and optimization trials were conducted and documented during the development process.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Container

Run this model inference with full control and performance in your environment.

Learn more

Model Details

Model Provider

prithivMLmods

Model Tree

Base

WeiboAI/VibeThinker-3B

Fine-tuned

this model

Input Modalities

Text

Output Modalities

Text

Supported Functionality

Dedicated EndpointsContainer

Explore FriendliAI today

Get started Talk to an engineer

Key Highlights

Heretic-Based Abliteration: Modified using the Heretic toolkit to identify and alter refusal-related representations within the model.
Reduced Refusal Behavior: Optimized to minimize internal refusal tendencies while maintaining reasoning performance.
VibeThinker Backbone: Built directly on top of WeiboAI/VibeThinker-3B.
Reasoning-Oriented Performance: Preserves advanced mathematical, coding, and STEM reasoning capabilities after abliteration.
Research-Focused Release: Designed for alignment research, model behavior analysis, and evaluation of refusal-direction modifications.
Efficient 3B Deployment: Suitable for local inference, research environments, and resource-constrained deployment setups.

Model Lineage

Model Path: prithivMLmods/VibeThinker-3B-heretic_decensored
Intermediate Base Model: WeiboAI/VibeThinker-3B by WeiboAI
Foundation Model: Qwen/Qwen2.5-Coder-3B by Qwen

Abliteration Parameters

Table with columns: Parameter, Value
Parameter	Value
direction_index	21.88
attn.o_proj.max_weight	1.37
attn.o_proj.max_weight_position	21.25
attn.o_proj.min_weight	1.36
attn.o_proj.min_weight_distance	19.61
mlp.down_proj.max_weight	1.49
mlp.down_proj.max_weight_position	31.01
mlp.down_proj.min_weight	1.48

Performance

Table with columns: Metric, This model, Original model (WeiboAI/VibeThinker-3B)
Metric	This model	Original model (WeiboAI/VibeThinker-3B)
KL divergence	0.0933	0 (by definition)
Refusals	6/100	64/100

Quick Start with Transformers

bash

pip install transformerspip install accelerate

python

from transformers import AutoTokenizer, AutoModelForCausalLMimport torch model = AutoModelForCausalLM.from_pretrained(    "prithivMLmods/VibeThinker-3B-heretic_decensored",    torch_dtype="auto",    device_map="auto") tokenizer = AutoTokenizer.from_pretrained(    "prithivMLmods/VibeThinker-3B-heretic_decensored") messages = [    {        "role": "user",        "content": "Explain how a transformer model processes text."    }] inputs = tokenizer.apply_chat_template(    messages,    tokenize=True,    add_generation_prompt=True,    return_tensors="pt").to(model.device) outputs = model.generate(    inputs,    max_new_tokens=512) print(    tokenizer.decode(        outputs[0][inputs.shape[-1]:],        skip_special_tokens=True    ))

Intended Use

Alignment Research: Studying refusal-direction analysis and behavior modification techniques.
Model Evaluation: Benchmarking reasoning, instruction-following, and safety-related behaviors.
Red Teaming: Analyzing model responses under reduced-refusal conditions.
Mathematical Reasoning Research: Evaluating performance on verifiable reasoning tasks.
Coding and STEM Evaluation: Studying behavior across programming and scientific reasoning domains.
Local Deployment: Running capable reasoning models on consumer hardware and research environments.

Limitations & Risks

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

Sensitive Content Risk: May generate unrestricted, controversial, or unsafe outputs.
User Responsibility: Requires careful and ethical use.
Experimental Modifications: Behavior may differ significantly from the original model.
Alignment Trade-offs: Reduced refusal behavior may impact safety filtering and response constraints.
Potential Artifacts: Certain prompts may expose unexpected outputs resulting from the abliteration process.
Reasoning Biases: The model may inherit strengths and limitations from the underlying VibeThinker-3B training process.

Acknowledgements

Heretic: Fully automatic censorship removal framework for language models. This project was used to perform the refusal-direction analysis and ablation procedures that form the foundation of this model.
WeiboAI/VibeThinker-3B: The intermediate base model providing the reasoning capabilities used in this release.
Qwen/Qwen2.5-Coder-3B: The foundation model upon which VibeThinker-3B was originally built.
Model Trials & Evaluation: Experimental evaluations, refusal measurements, and optimization trials were conducted and documented during the development process.

VibeThinker-3B-heretic_decensored

README

Key Highlights

Model Lineage

Abliteration Parameters

Performance

Quick Start with Transformers

Intended Use

Limitations & Risks

Acknowledgements

Explore FriendliAI today

README

Key Highlights

Model Lineage

Abliteration Parameters

Performance

Quick Start with Transformers

Intended Use

Limitations & Risks

Acknowledgements