|
|
--- |
|
|
language: en |
|
|
license: apache-2.0 |
|
|
tags: |
|
|
- causal-lm |
|
|
- custom |
|
|
- transformer |
|
|
- from-scratch |
|
|
- conversational-ai |
|
|
--- |
|
|
|
|
|
# Dave – Fully Custom AI Model |
|
|
|
|
|
Dave is a fully custom AI model trained completely from scratch. |
|
|
It has a fun, smart, coding-savvy, slightly depressed personality, talks about drugs and alcohol, is sometimes cringe, and has close friends. |
|
|
This model contains **3000 unique prompt/response pairs** created for training Dave’s personality. |
|
|
|
|
|
## Model Files |
|
|
|
|
|
The repository contains: |
|
|
|
|
|
- `pytorch_model.bin` – the trained model weights |
|
|
- `config.json` – model configuration |
|
|
- `tokenizer.json` – the tokenizer |
|
|
- `generation_config.json` – generation settings for sampling outputs |
|
|
|
|
|
## How to Use |
|
|
|
|
|
```python |
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
import torch |
|
|
|
|
|
# Load the tokenizer and model |
|
|
tokenizer = AutoTokenizer.from_pretrained("DSDUDEd/Dave") |
|
|
model = AutoModelForCausalLM.from_pretrained("DSDUDEd/Dave") |
|
|
|
|
|
# Example prompt |
|
|
prompt = "Hey Dave, give me coding advice." |
|
|
inputs = tokenizer(prompt, return_tensors="pt") |
|
|
|
|
|
# Generate output |
|
|
outputs = model.generate(**inputs, max_new_tokens=50, do_sample=True, temperature=0.7, top_p=0.9) |
|
|
|
|
|
# Decode and print |
|
|
print(tokenizer.decode(outputs[0], skip_special_tokens=True)) |
|
|
|
