Instructions to use hherb/AstaBrief-8B-MLX-8bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hherb/AstaBrief-8B-MLX-8bit with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("hherb/AstaBrief-8B-MLX-8bit")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

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

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use hherb/AstaBrief-8B-MLX-8bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "hherb/AstaBrief-8B-MLX-8bit"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "hherb/AstaBrief-8B-MLX-8bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use hherb/AstaBrief-8B-MLX-8bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "hherb/AstaBrief-8B-MLX-8bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "hherb/AstaBrief-8B-MLX-8bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "hherb/AstaBrief-8B-MLX-8bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use hherb/AstaBrief-8B-MLX-8bit with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "hherb/AstaBrief-8B-MLX-8bit"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default hherb/AstaBrief-8B-MLX-8bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use hherb/AstaBrief-8B-MLX-8bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "hherb/AstaBrief-8B-MLX-8bit"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "hherb/AstaBrief-8B-MLX-8bit" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

AstaBrief 8B — MLX 8-bit

An 8-bit MLX conversion of Ai2's AstaBrief 8B, a Qwen3-based model that generates cited research reports from a question and supplied literature excerpts.

Converted with MLX-LM using 8-bit affine quantization, group size 64, and BF16 for non-quantized parameters. The two weight shards total approximately 8.70 GB (8.11 GiB).

Source revision: 6a34f54dcd7887b790a570249b4a2a2fe81c6694. All four source shard SHA-256 checksums were verified against Hugging Face repository metadata before conversion. Package versions, quantization settings and output weight checksums are recorded in conversion-info.json.

Use

pip install --upgrade mlx-lm

mlx_lm.generate \
  --model hherb/AstaBrief-8B-MLX-8bit \
  --prompt "What can you help me with?" \
  --max-tokens 256

Use an Apple silicon Mac with enough unified memory for the model, context cache and application overhead. Longer contexts require additional memory.

For research report generation, populate Ai2's recommended prompt by replacing [QUERY] and [SECTION_REFERENCES] with your question and retrieved literature excerpts. Save the populated text as prompt.txt, then run:

mlx_lm.generate \
  --model hherb/AstaBrief-8B-MLX-8bit \
  --prompt - \
  --temp 0.7 \
  --top-p 0.95 \
  --max-tokens 4096 < prompt.txt

The source tokenizer and chat template are included, and MLX-LM applies the chat template by default. Literature retrieval must be supplied separately.

Conversion

The source repository's safetensors filenames were normalized locally from pytorch_model-*.safetensors to model-*.safetensors for the MLX-LM loader, without changing their contents. The conversion command was:

mlx_lm.convert \
  --hf-path ./AstaBrief-source \
  --mlx-path ./AstaBrief-8B-MLX-8bit \
  --quantize --q-bits 8 --q-group-size 64 \
  --dtype bfloat16

Validation

The saved quantized model reloaded successfully and generated 768 tokens using the recommended prompt format and two synthetic caching excerpts. The response included both supplied reference IDs. smoke-test.txt contains that response; it stops at the test's token limit.

This is a loading and generation check, not a research-quality benchmark or an evaluation of equivalence to the original BF16 checkpoint. The original model's published benchmark scores do not establish scores for this quantized conversion.

Attribution and license

Original model and training: Ai2 / allenai. This repository provides a format conversion and quantization of their checkpoint, without additional training.

The weights retain the source model's Apache 2.0 license; a copy is included in LICENSE. See the preserved source model card and original repository for intended use, limitations, training details and upstream attribution. The linked prompt dataset has its own license.