Ishaank18/cmrl-adaptive-generator

CMRL Adaptive Generator

Multi-level adaptive generator trained with SFT for the C-MRL project.

Model Description

Adapts explanations to different difficulty levels:

• Novice: 6th grade level, simple analogies
• Intermediate: College student level
• Expert: Technical/professional explanations

Training Details

Parameter	Value
Base Model	unsloth/Qwen2-1.5B-Instruct-bnb-4bit
LoRA Rank	16
LoRA Alpha	32
Target Modules	q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
Epochs	3
Learning Rate	0.0001
Final Train Loss	1.5193
Final Eval Loss	0.0000

Team

Team kats - IIIT Hyderabad

Usage

from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    "Ishaank18/cmrl-adaptive-generator",
    max_seq_length=2048,
    load_in_4bit=True,
)
FastLanguageModel.for_inference(model)

messages = [
    {"role": "system", "content": "You are an adaptive tutor."},
    {"role": "user", "content": "Explain photosynthesis to a 6th grader."}
]
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.7)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))