tinyllama-erp-merged

This model is a fine-tuned and merged version of TinyLlama/TinyLlama-1.1B-Chat-v1.0 for ERP-related question answering and text generation.

Model details

• Base model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
• Model type: Causal language model
• Framework: Hugging Face Transformers
• Format: Safetensors
• Use case: ERP question answering and assistance

Intended use

This model is intended for:

• ERP-related question answering
• General business process assistance

Limitations

• It may produce incomplete or incorrect answers.
• It should not be used for critical financial or legal decisions without human review.
• Output quality depends on the training data.

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

repo_id = "nairanu6115/tinyllama-erp-merged"

tokenizer = AutoTokenizer.from_pretrained(repo_id)
model = AutoModelForCausalLM.from_pretrained(
    repo_id,
    torch_dtype=torch.float16,
    device_map="auto"
)

prompt = "What is a purchase order?"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

with torch.no_grad():
    outputs = model.generate(
        **inputs,
        max_new_tokens=200,
        do_sample=False,
        pad_token_id=tokenizer.eos_token_id
    )

print(tokenizer.decode(outputs[0], skip_special_tokens=True))