Llama-3.2-3B-Thai_CRM Specialist
Fine-tuned version of meta-llama/Llama-3.2-3B optimized for Customer Relationship Management (CRM), Customer Behavior Analysis, and Thai business workflows.
Key Capabilities
- CRM Chatbot & Agent: Context-aware customer service interactions and automated ticket handling in Thai & English.
- CRM Analyzer: Evaluates purchase history, sentiment, and customer data to suggest retention strategies.
Model Description
- Developed by: yothinS
- Model type: Causal Language Model
- Language(s): Thai, English
- License: Llama 3.2 Community License
- Base model: meta-llama/Llama-3.2-3B
Training Procedure
Trained using Unsloth and Hugging Face's TRL library.
- Method: LoRA (Rank: 16, Alpha: 32)
- Dataset:
yothinS/Customer_Behavior_Analysis(~X,000 samples) - Epochs: 3
- Precision: 16-bit (bfloat16)
Intended Uses & Limitations
- Best for: CRM query routing, customer feedback analysis, retention campaign drafting.
- Limitations: Outputs require human oversight for sensitive billing or legal disputes. May hallucinate outside the CRM/behavioral analytics domain.
How to Use
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
model_id = "yothinS/Llama-3.2-3B-Thai_CRM"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.bfloat16,
device_map="auto"
)
messages = [
{"role": "system", "content": "You are a professional CRM and customer behavior specialist."},
{"role": "user", "content": "วิเคราะห์พฤติกรรมลูกค้ารายนี้และแนะนำโปรโมชันรักษาฐานลูกค้า: ยอดซื้อเฉลี่ยลดลง 40% ในช่วง 3 เดือนที่ผ่านมา"}
]
inputs = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt").to("cuda")
outputs = model.generate(inputs, max_new_tokens=300, temperature=0.7)
print(tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True))