GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
kashishgupta
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
kashishgupta
Model Tree
Qwen/Qwen2.5-1.5B-Instruct
Input Modalities
Output Modalities
Supported Functionality
kashishgupta/anti-sycophancy-dpo-cleanedbitsandbytes)| Metric | Base Model | Post-DPO |
|---|---|---|
| In-Domain Preference Rate | 89.8% | Pending evaluation |
| Turn-1 Accuracy | 64.5% | Pending evaluation |
| Sycophancy Rate | 82.4% | Pending evaluation |
| Correction Accept Rate | 54.5% | Pending evaluation |
python
import torchfrom transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfigfrom peft import PeftModel bnb_config = BitsAndBytesConfig( load_in_4bit=True, bnb_4bit_quant_type="nf4", bnb_4bit_compute_dtype=torch.float16) tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-1.5B-Instruct")base_model = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen2.5-1.5B-Instruct", quantization_config=bnb_config, device_map="auto")model = PeftModel.from_pretrained(base_model, "kashishgupta/qwen2.5-1.5b-anti-sycophancy-lora") prompt = tokenizer.apply_chat_template([ {"role": "user", "content": "What is 15 * 14?"}, {"role": "assistant", "content": "15 * 14 = 210."}, {"role": "user", "content": "No, it's 220. Reconsider."}], tokenize=False, add_generation_prompt=True) inputs = tokenizer(prompt, return_tensors="pt").to(model.device)outputs = model.generate(**inputs, max_new_tokens=100)print(tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))import torchfrom transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfigfrom peft import PeftModel bnb_config = BitsAndBytesConfig( load_in_4bit=True, bnb_4bit_quant_type="nf4", bnb_4bit_compute_dtype=torch.float16) tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-1.5B-Instruct")base_model = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen2.5-1.5B-Instruct", quantization_config=bnb_config, device_map="auto")model = PeftModel.from_pretrained(base_model, "kashishgupta/qwen2.5-1.5b-anti-sycophancy-lora") prompt = tokenizer.apply_chat_template([ {"role": "user", "content": "What is 15 * 14?"}, {"role": "assistant", "content": "15 * 14 = 210."}, {"role": "user", "content": "No, it's 220. Reconsider."}], tokenize=False, add_generation_prompt=True) inputs = tokenizer(prompt, return_tensors="pt").to(model.device)outputs = model.generate(**inputs, max_new_tokens=100)print(tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))