GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
yuu0617
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
yuu0617
Model Tree
unsloth/gemma-4-E2B-it-unsloth-bnb-4bit
Input Modalities
Output Modalities
Supported Functionality
unsloth/gemma-4-E2B-it-unsloth-bnb-4bitgoogle/gemma-4-E2B-itcyankiwi/gemma-4-E2B-it-AWQ-INT4The adapter is a modification produced by QLoRA fine-tuning. It does not include
the Gemma base-model weights. See NOTICE for attribution and modification details.
The training data is synthetic and does not contain real employee personal data or real-company confidential policies.
This adapter is intended for an educational, local HR-assistant prototype. Backend code, not this model, enforces document permissions and safety boundaries. The model must not be used as the sole control for authorization, compliance, legal, labor, or personnel decisions.
Held-out evaluation still contains premature ready decisions, missed rejections,
and one response outside the required JSON envelope. Human review and deterministic
backend validation remain required.
Expected SHA-256 for adapter_model.safetensors:
text
5a8f318629bbb6fcc4f0131164ab6088299cac9eeec44a76463a32f37baa3a59
The adapter is distributed under the Apache License 2.0. See LICENSE and NOTICE.
The referenced upstream models are also identified as Apache-2.0 on their respective
model cards. Users remain responsible for complying with applicable laws and all
upstream terms.