YoAbriel/KodaLite-1.3B-mlx

KodaLite-1.3B — MLX (fp16)

MLX version of YoAbriel/KodaLite-1.3B, optimized for Apple Silicon (M1/M2/M3/M4).

Size: ~2.5 GB | Precision: bfloat16

Usage

pip install mlx-lm

from mlx_lm import load, generate

model, tok = load("YoAbriel/KodaLite-1.3B-mlx")
prompt = tok.apply_chat_template(
    [{"role": "user", "content": "What is the capital of France?"}],
    tokenize=False,
    add_generation_prompt=True,
)
print(generate(model, tok, prompt=prompt, max_tokens=80))

Or from the command line:

mlx_lm.generate --model YoAbriel/KodaLite-1.3B-mlx \
  --prompt "<|user|>\nHello\n<|assistant|>\n" --max-tokens 80

Other quantizations

• YoAbriel/KodaLite-1.3B-mlx-8bit — 1.4 GB, 8-bit

Limitations

Small model (1.27B params), undertrained (1.64B tokens). See the base model card for full details.

License

Apache 2.0