Gemma-4-31B-Humanlike-Chat (FP8)

A 31B Gemma that texts like a person and still does the work, in FP8. No system prompt needed.

Results, examples and how it was made are on the main card: Gemma-4-31B-Humanlike-Chat-GGUF.

"Base" here always means Google's google/gemma-4-31B-it, the official release. It is not abliterated.

vLLM

Tested with vLLM 0.31.0 on an RTX PRO 6000 Blackwell (96 GB):

vllm serve LessThanThreeAI/Gemma-4-31B-Humanlike-Chat-FP8 --served-model-name humanlike-gemma \
  --max-model-len 16384 --language-model-only \
  --enable-auto-tool-choice --tool-call-parser gemma4 --reasoning-parser gemma4

Thinking is off by default, as in Google's Gemma 4 template. Send chat_template_kwargs: {"enable_thinking": true} per request to turn it on; the thinking comes back in its own field. It was trained with effort levels: add the line Reasoning effort: low or Reasoning effort: high to your system prompt, or no line for medium. Sampling: temperature 1.0, top_p 0.95, top_k 64 (Google's defaults in generation_config.json). For SGLang, point --model-path at this repo.

Format

FP8 weights (per channel) with dynamic per-token FP8 activations (FP8_DYNAMIC), made with llm-compressor 0.14.0.1 in compressed-tensors format. All 410 language linear layers are FP8; the vision tower, embeddings and lm_head stay BF16. Made from the BF16 weights.

VRAM

The weights take 30.4 GiB in vLLM. On the 96 GB test card at --gpu-memory-utilization 0.88, that left 50.0 GiB for the cache, 59,496 tokens. With less memory, take the GPTQ-Int4 repo or a GGUF from the main card.

Checks

Same check as every file on the main card: 5 fresh test chats against the reference (the merged BF16 weights in vLLM, which match Google's base plus the Humanlike adapter at runtime: KL 0.0005), served on vLLM 0.31.0, one NVIDIA RTX PRO 6000 Blackwell Server Edition (96 GB).

Commissions

Commissions open, DM codebottle on Discord (https://discord.com/users/320486798859960322). I build custom finetunes like this one: characters, product voices, distillation into smaller models, and domain or use-case specific models.

License

Apache-2.0, inherited from Google's Gemma 4 release.