tpls

gemma-4-12B-coder-fable5-composer2.5-v1-sft-v5-abliterated

Dedicated Endpoints

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

Get help setting up a custom Dedicated Endpoints.

Talk with our engineer to get a quote for reserved GPU instances with discounts.

Abliterated (refusal-direction-ablated, KL-bounded) derivative of tpls/gemma-4-12B-coder-fable5-composer2.5-v1-sft-v5, via Heretic with KL_MAX=0.2.

bf16 safetensors at the repo root.
imatrix GGUF quants under gguf/ (IQ4_XS Q3_K_M Q4_K_M Q5_K_M Q6_K Q8_0), gemma-4 EOG-patched (eot=106/eom=105) so generation stops correctly.

Model provider

tpls

Model tree

Base

tpls/gemma-4-12B-coder-fable5-composer2.5-v1-sft-v5

Fine-tuned

this model

Modalities

Input

Video, Audio, Text, Image

Output

Text