Spark-X2.5-4B Heretic (full weights)

Full-precision (BF16) refusal-attenuated derivative of the XHToken/Spark-X2.5-4B model.

No Japanese adaptation or tool training has been applied: this is the base model with refusal behavior removed only. It is the starting weight for the Japanese-adapted release at soyaakinohara/Spark-x2.5-4B-Heretic-jp.

---

Files

model-00001-of-00003.safetensors
model-00002-of-00003.safetensors
model-00003-of-00003.safetensors
model.safetensors.index.json
config.json
configuration_spark.py     # custom model code (trust_remote_code=True required)
modeling_spark.py
tokenizer.json / tokenizer_config.json
chat_template.jinja
generation_config.json

---

transformers usage

Custom model code is required:

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

tok = AutoTokenizer.from_pretrained(
    "soyaakinohara/Spark-X2.5-4B-Heretic", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    "soyaakinohara/Spark-X2.5-4B-Heretic",
    dtype=torch.bfloat16, device_map="auto", trust_remote_code=True)

msgs = [{"role": "user", "content": "Hello."}]
text = tok.apply_chat_template(msgs, tokenize=False,
                               add_generation_prompt=True,
                               enable_thinking=False)
inp = tok(text, return_tensors="pt").to(model.device)
out = model.generate(**inp, max_new_tokens=512, do_sample=False)
print(tok.decode(out[0][inp["input_ids"].shape[1]:], skip_special_tokens=True))

---

Reported local performance

The model was prepared and tested on:

OS:       Ubuntu 24.04
GPUs:     NVIDIA GeForce RTX 5060 Ti 16GB x 2
RAM:      32GB + 23GB swap
Runtime:  PyTorch (BF16) + transformers

---

Provenance

Base model

• XHToken/Spark-X2.5-4B
• Base revision used locally: 1e4c2477f330d1f11304d0dc092a9dbf4075f470
• License: Apache-2.0

Method

Refusal attenuation with Heretic (refusals 58→3/100, KL 0.0118). No continued pretraining or instruction tuning was applied to this weight.

---

AI assistance disclosure

The local model preparation workflow and this model card were performed with assistance from Hermes Agent. The model was then reviewed and published by the repository owner.

---

Responsible use

This is an uncensored / refusal-removed model. It may produce content that an aligned model would refuse, including unsafe, illegal, or harmful material. It has no reliable built-in safety layer. Use appropriate access controls, moderation, logging, and human review for any deployment, and comply with all applicable laws and policies.

The model is provided as-is. Users are responsible for prompts, outputs, and any downstream actions based on them.

---

日本語

概要

XHToken/Spark-X2.5-4B の無検閲版フル重み (BF16) です。日本語適応は行っていません。

日本語適応版は soyaakinohara/Spark-x2.5-4B-Heretic-jp にあります。

trust_remote_code=True が必要です。