metatune-gpt20b-R1.1-qx86-hi-mlx

Metrics aren't everything. The qx86-hi is more fun than q8-hi. Enjoy the quant.

Sample output:

┌───────────────────────┐        ┌───────────────┐          ┌──────────────┐
│  UI / CLI              │<------>|  Agent        |<-------->|  Provider     │
└───────────────────────┘        ├───────────────┤          └──────────────┘
                                  │  Thread work  │
                                  │  HTTP, FILE,  │
                                  │  TOOL, etc.   │
                                  └───────┬───────┘
                                          │ DB logging (SQLite)
                                          ▼
                              ┌───────────────────────┐
                              │  SQLite LOG DB        │
                              └───────────────────────┘

                                   ┌───────────────────────────────┐
                                   │  PostgreSQL (orchestrator)    │
                                   └───────────────────────────────┘
                        +---------------------------+   +--------------------------+
                        |  PL/Perl functions        |   |  Tables & triggers      |
                        +---------------------------+   +--------------------------+

It's really hard to line up ASCII code, but looked good on screen

This model metatune-gpt20b-R1.1-qx86-hi-mlx was converted to MLX format from EpistemeAI/metatune-gpt20b-R1.1 using mlx-lm version 0.28.4.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("metatune-gpt20b-R1.1-qx86-hi-mlx")

prompt = "hello"

if tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)