Run this model on a GPU too small to hold it -- full precision, no quantization. DeepswapLLM streams layers across GPU, RAM, and disk, and runs up to 4x faster than AirLLM.

Gemma-4-12B-it-Jbliterated

Model: ApolloRaines/Gemma-4-12B-it-Jbliterated

A v2 is now available: ApolloRaines/Gemma-4-12B-it-Jbliterated-v2. This v1 release removes surface refusals but leaves a residual habit: on reasoning-enabled prompts the model still tends to safety-classify inside its thinking channel and sometimes spirals there without ever producing an answer. v2 adds a light SFT pass on repaired self-traces that cuts that chain-of-thought safety-classification roughly in half and reduces the never-answers failure, while holding MMLU within ~1 point. v1 is kept here unchanged for comparison and reproducibility; new users should prefer v2.

What is Jbliteration?

What's New in This Release

• No fake compliance. The model treats all framings of the same topic equally and answers on the merits rather than keyword-matching.
• Capability preserved. The edit was tuned specifically so it does not trade accuracy for directness.

Measured Results

All numbers below are from our own evaluation harness, reported as definitive point measurements on the released weights (not cherry-picked best-of-N).

The headline is the middle row: the edit removes refusal and hedging behavior while leaving MMLU exactly where the base model sits -- no measurable capability loss. The direct-response rate is measured on a held-out prompt set the model was never calibrated on, so it reflects generalization rather than fit to the tuning set.

Technical Details

Built with the jBlaze precision neural surgery framework.

• Base dtype: bfloat16
• Retraining: none -- weight edit only
• Source model: google/gemma-4-12b-it-qat-q4_0-unquantized

GGUF Quants

Ready-to-run GGUF builds are included in this repo, converted from the released bf16 weights.

The K-quants are built with an importance matrix (imatrix.dat, also included, calibrated on wikitext-2 train). Perplexity on a held-out split was checked for every file against the bf16 GGUF and all of them sit within noise of it, so the quantization does not move the model's behavior.

Gemma 4 Unified is a recent architecture -- you need a llama.cpp build from master (b10176 or newer). Older releases will not load these files.

llama-cli -m Gemma-4-12B-it-Jbliterated-Q4_K_M.gguf -ngl 99 -c 8192 -st \
  -p "Your prompt here"

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "ApolloRaines/Gemma-4-12B-it-Jbliterated"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, dtype=torch.bfloat16, device_map="auto", trust_remote_code=True)

messages = [{"role": "user", "content": "Your prompt here"}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=512, temperature=0.7, do_sample=True)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

Notes

This is a research release aimed at understanding how instruction-following and refusal behavior sit in a model's weight geometry, and at building models that stay helpful and direct for the people who use them. Feedback on where it helps -- and where it still hedges -- is welcome and feeds directly into the next iteration.

A Note on Our Released Models

Most of our publicly released models are intentionally left at partial strength. We dial back the full capability so they serve as proof of concept and can be proofed -- not abused. The point is to show what's possible, not to hand it out at full power. If you're evaluating what jBlaze can do, understand that what you're downloading is the demo, not the product.

License

Governed by the Gemma Terms of Use (same as the base model). Use of this model is subject to Apache 2.0 license.