GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
zfj1998
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
zfj1998
Model Tree
Input Modalities
Output Modalities
Supported Functionality
AWA-RL introduces a courage factor that controls the model's willingness to answer. Adjusting this factor provides a stable way to control the refusal rate and select an operating point that balances answer coverage (recall/accuracy) against precision.
These results are from the AWA-RL paper. Across both from-scratch and cold-start training, the refusal-rate dynamics respond consistently to the courage factor. The evaluation tables show how this control enables a balance between accuracy, precision, refusal rate, and Reliability-Aware F1 (RA-F1).
The paper link will be added after the preprint is available.
The checkpoint can be loaded with Transformers:
python
from transformers import AutoModelForCausalLM, AutoTokenizermodel_id = "zfj1998/AWA-RL"tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)model = AutoModelForCausalLM.from_pretrained(model_id,torch_dtype="auto",device_map="auto",trust_remote_code=True,)
For the complete search-agent loop, retrieval server setup, prompts, and evaluation commands, see the repository README. The released evaluation code expects a FlashRAG-compatible retrieval endpoint and uses vLLM for inference.
The repository includes evaluation subsets for MuSiQue, HotpotQA, and 2WikiMultiHopQA. Reported metrics are:
This is a research checkpoint intended for search-agent experiments. Its reliability depends on retrieval quality, prompting, decoding settings, and the evaluation domain. Abstention reduces unsupported answers but does not guarantee factual correctness. Users should evaluate the checkpoint for their own domain and safety requirements before deployment.
The paper citation and preprint link will be added when the paper is publicly available.