Playdate DistilGPT2: 5k Samples

Expect this to be the last model based on DistilGPT2 as coherence has not advanced. Test model for newer personalities, not to be used in production. Script example below.

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

MODEL_PATH = "x5tne/playdate-distilgpt2-5k" 

# Load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH)
model = AutoModelForCausalLM.from_pretrained(MODEL_PATH)
model.eval()

# Example prompt
prompt = """<system harsh>
<summary>none</summary>
<user>Hi [namehere]! How are you today?</user>
<assistant>"""

# Encode input
inputs = tokenizer(prompt, return_tensors="pt")

# Generate output
with torch.no_grad():
    outputs = model.generate(
        **inputs,
        max_new_tokens=100,
        do_sample=True,
        temperature=0.9,
        top_p=0.9,
        repetition_penalty=1.15
    )

# Decode generated tokens
response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)

print("Assistant response:")
print(response)