GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
shuvo-halder
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
shuvo-halder
Model Tree
Qwen/Qwen2.5-1.5B-Instruct
Input Modalities
Output Modalities
Supported Functionality
bn), English (en)Qwen/Qwen2.5-1.5B-InstructTo prevent resource exhaustion and session timeouts on free-tier cloud GPUs (Google Colab), the training was executed in two sequential stages across the full 51,760 samples:
| Stage | Dataset Split | Examples | Learning Rate | Status |
|---|---|---|---|---|
| Stage 1 (Initial 50%) | train[:50%] | ~25,880 | 2e-4 | Completed |
| Stage 2 (Final 50%) | train[50%:] | ~25,880 | 1e-4 | Completed |
1e-4 during Stage 2, the model assimilated the remainder of the dataset without experiencing catastrophic forgetting of the first half.Use the snippet below to load and run inference with 🤗 transformers and peft:
python
import torchfrom peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizer base_model_id = "Qwen/Qwen2.5-1.5B-Instruct"adapter_id = "shuvo-halder/Qwen2.5-1.5B-Bangla" # ১. টোকেনাইজার এবং বেস মডেল লোডtokenizer = AutoTokenizer.from_pretrained(base_model_id)base_model = AutoModelForCausalLM.from_pretrained( base_model_id, torch_dtype=torch.float16, device_map="auto") # ২. ফাইন-টিউনড LoRA অ্যাডাপ্টার লোড করাmodel = PeftModel.from_pretrained(base_model, adapter_id)model.eval() # ৩. বাংলায় চ্যাট প্রম্পট তৈরিuser_question = "নিয়মিত বই পড়ার উপকারিতা কী কী?" messages = [{"role": "user", "content": user_question}]prompt = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True) inputs = tokenizer(prompt, return_tensors="pt").to("cuda") # ৪. উত্তর জেনারেট করাwith torch.no_grad(): outputs = model.generate( **inputs, max_new_tokens=256, temperature=0.7, top_p=0.9, repetition_penalty=1.1, do_sample=True, pad_token_id=tokenizer.eos_token_id ) input_length = inputs["input_ids"].shape[1]response = tokenizer.decode(outputs[0][input_length:], skip_special_tokens=True) print("উত্তর:\n", response.strip())import torchfrom peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizer base_model_id = "Qwen/Qwen2.5-1.5B-Instruct"adapter_id = "shuvo-halder/Qwen2.5-1.5B-Bangla" # ১. টোকেনাইজার এবং বেস মডেল লোডtokenizer = AutoTokenizer.from_pretrained(base_model_id)base_model = AutoModelForCausalLM.from_pretrained( base_model_id, torch_dtype=torch.float16, device_map="auto") # ২. ফাইন-টিউনড LoRA অ্যাডাপ্টার লোড করাmodel = PeftModel.from_pretrained(base_model, adapter_id)model.eval() # ৩. বাংলায় চ্যাট প্রম্পট তৈরিuser_question = "নিয়মিত বই পড়ার উপকারিতা কী কী?" messages = [{"role": "user", "content": user_question}]prompt = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True) inputs = tokenizer(prompt, return_tensors="pt").to("cuda") # ৪. উত্তর জেনারেট করাwith torch.no_grad(): outputs = model.generate( **inputs, max_new_tokens=256, temperature=0.7, top_p=0.9, repetition_penalty=1.1, do_sample=True, pad_token_id=tokenizer.eos_token_id ) input_length = inputs["input_ids"].shape[1]response = tokenizer.decode(outputs[0][input_length:], skip_special_tokens=True) print("উত্তর:\n", response.strip())