GLM-4.7-Flash-heretic

Abliterated (uncensored) version of zai-org/GLM-4.7-Flash, created using Heretic and converted to GGUF.

Abliteration Quality

Metric	Value
Refusals	6/100
KL Divergence	0.0071
Rounds	1

Lower refusals = fewer refused prompts. Lower KL divergence = closer to original model behavior.

Available Quantizations

Quantization	File	Size
Q8_0	GLM-4.7-Flash-heretic-Q8_0.gguf	31.80 GB
Q6_K	GLM-4.7-Flash-heretic-Q6_K.gguf	22.92 GB
Q4_K_M	GLM-4.7-Flash-heretic-Q4_K_M.gguf	16.89 GB

Usage with Ollama

ollama run hf.co/ThalisAI/GLM-4.7-Flash-heretic:Q8_0
ollama run hf.co/ThalisAI/GLM-4.7-Flash-heretic:Q6_K
ollama run hf.co/ThalisAI/GLM-4.7-Flash-heretic:Q4_K_M

bf16 Weights

The full bf16 abliterated weights are available in the bf16/ subdirectory of this repository.

Usage with Transformers

The bf16 weights in the bf16/ subdirectory can be loaded directly with Transformers:

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "ThalisAI/GLM-4.7-Flash-heretic"
tokenizer = AutoTokenizer.from_pretrained(model_id, subfolder="bf16")
model = AutoModelForCausalLM.from_pretrained(
    model_id, subfolder="bf16", torch_dtype="auto", device_map="auto"
)

messages = [{"role": "user", "content": "Hello!"}]
text = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))

About

This model was processed by the Apostate automated abliteration pipeline:

1. The source model was loaded in bf16
2. Heretic optimization-based abliteration was applied to remove refusal behavior
3. The merged model was converted to GGUF format using llama.cpp
4. Multiple quantization levels were generated

The abliteration process uses directional ablation to remove the model refusal directions while minimizing KL divergence from the original model behavior on harmless prompts.