GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
ehzawad
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
ehzawad
Model Tree
Base
Qwen/Qwen3-1.7B
Adapter
this model
Input Modalities
Text
Output Modalities
Text
Supported Functionality
Dedicated Endpoints
Container
Qwen/Qwen3-1.7BTo use this adapter, load the base model and apply the PEFT adapter:
python
!pip install --upgrade torchaoimport torchfrom transformers import AutoModelForCausalLM, AutoTokenizerfrom peft import PeftModelbase_model_id = "Qwen/Qwen3-1.7B"adapter_id = "ehzawad/qwen3_1_7b-gsm8k-grpo"# 1. Load tokenizertokenizer = AutoTokenizer.from_pretrained(adapter_id)if tokenizer.pad_token is None:tokenizer.pad_token = tokenizer.eos_tokentokenizer.padding_side = "left"# 2. Load base modelmodel = AutoModelForCausalLM.from_pretrained(base_model_id,torch_dtype=torch.bfloat16,device_map="auto")# 3. Load and apply adaptermodel = PeftModel.from_pretrained(model, adapter_id)model.eval()# 4. Prepare promptsystem_prompt = "You are a careful math reasoning assistant. Solve the problem step by step, but keep the solution concise. Use only the needed calculations, avoid repetition, and end with exactly one final answer in the form \\boxed{answer}."question = "Janet has 3 bags with 4 apples each. She gives away 5 apples and then took back 4. Then ate 3 apples and then friends took away 2 apples and then he boughts 5 apples again. How many remain?"messages = [{"role": "system", "content": system_prompt},{"role": "user", "content": question}]# 5. Format and Generatetry:inputs = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, enable_thinking=True, return_dict=True, return_tensors="pt").to(model.device)except TypeError:inputs = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_dict=True, return_tensors="pt").to(model.device)with torch.inference_mode():outputs = model.generate(**inputs,max_new_tokens=512,do_sample=True,temperature=0.6,pad_token_id=tokenizer.pad_token_id,eos_token_id=tokenizer.eos_token_id)print(tokenizer.decode(outputs[0][inputs['input_ids'].shape[-1]:], skip_special_tokens=True))