from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
from peft import PeftModel

# Load model
tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-base")
base_model = AutoModelForSeq2SeqLM.from_pretrained("google/flan-t5-base")
model = PeftModel.from_pretrained(base_model, "YOUR_USERNAME/flan-t5-conversation-generator")

# Generate
prefix = "Generate a call center conversation from this summary: "
summary = "The client called about a broken AC unit. The agent scheduled a technician."

inputs = tokenizer(prefix + summary, return_tensors="pt", max_length=256, truncation=True)
outputs = model.generate(**inputs, max_new_tokens=512, num_beams=4)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Training Details

LoRA rank: 32
LoRA alpha: 64
Target modules: q, k, v, o
Epochs: 5
Learning rate: 5e-4

Example

Input Summary:

The client called about an ongoing issue where the heater making burning smell. The agent scheduled a technician to inspect heating elements.

Generated Conversation:

Client: Hi, I'm calling because my heater making burning smell.
Agent: Thanks for explaining. When did you first notice this happening?
Client: It started a couple days ago and keeps repeating.
Agent: Understood, we'll inspect heating elements during the service visit.

GoshKolotyan/flan-t5-conversation-generator

FLAN-T5 Conversation Generator (LoRA)

Model Description

Usage

Training Details

Example

GoshKolotyan/
flan-t5-conversation-generator