theguywhosucks/mochaV1-mini

Mocha Mini ☕

Mocha Mini is a tiny but mighty sentence completion model designed for fast, creative text generation without the bloat. Perfect for small projects, prototyping, or just having fun with text.

---

⚡ Features

• 🏎 Ultra-Lightweight – fast loading, easy deployment.
• ✍️ Sentence Completion – generate relevant endings from any text prompt.
• 🛡 Safetensors Format – secure, efficient, and easy to load.
• 🎨 Custom Tokenization – uses stoi/itos for fully local control.

---

📂 Project Structure

.
├── mocha.safetensors # Model weights
├── config.json
├── generation_config.json
├── itos.json         # Index-to-string mapping
├── stoi.json         # String-to-index mapping
├── logo.png          # Project logo
├── banner.png        # Project banner
└── README.md

---

🚀 Quick Start

No official tokenizer? No problem — Mocha Mini just needs your stoi/itos for input/output mapping:

import torch
from transformers import AutoModelForCausalLM, AutoConfig

# Load config and model
config = AutoConfig.from_pretrained("theguywhosucks/mochaV1-base", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    "theguywhosucks/mochaV1-base",
    config=config,
    device_map="auto",
    load_in_8bit=True,  # INT8 quantization for speed & memory
    trust_remote_code=True,
    use_safetensors=True
)

# Use your custom stoi/itos for input/output
def stoi(text): ...
def tosi(indices): ...

prompt = "The future of AI is"
tokens = stoi(prompt)
# Model inference logic here

---

📸 Assets

---

🛠 Requirements

• Python 3.9+
• Transformers
• Safetensors
• PyTorch
• bitsandbytes (optional for INT8 quantization)

Install all dependencies:

pip install torch transformers safetensors bitsandbytes

---

📜 License

Licensed under the Mocha Proprietary License — usage, distribution, and modification are restricted. See LICENSE for details.

---

☕ Mocha Mini – tiny, fast, and ready to complete your sentences.