GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
Jnx03
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
Jnx03
Model Tree
deepseek-ai/DeepSeek-R1-Distill-Qwen-14B
Input Modalities
Output Modalities
Supported Functionality
json
{"lr": 1e-05,"epochs": 1,"r": 16,"qlora": true,"mix": "30:70","bs": 4,"ga": 2,"max_len": 1536}
| Benchmark | Score | Target | Beat? |
|---|---|---|---|
| thaiexam | 35.93 | 75.0 | ❌ |
| math500 | 54.80 | 82.0 | ❌ |
| aime24 | 23.33 | 25.0 | ❌ |
python
from peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizerbase = "deepseek-ai/DeepSeek-R1-Distill-Qwen-14B"adapter = "Jnx03/kanitakorn-r1d-qwen14b-stage1-v1b-20260521"tok = AutoTokenizer.from_pretrained(base, trust_remote_code=True)model = AutoModelForCausalLM.from_pretrained(base, torch_dtype="bfloat16", device_map="auto", trust_remote_code=True)model = PeftModel.from_pretrained(model, adapter)
See VERSION_LOG.md for context on this checkpoint vs siblings.
License: MIT (inherited from deepseek-ai/DeepSeek-R1-Distill-Qwen-14B).