Instructions to use mlx-community/LongCat-Flash-Lite-Sparse-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mlx-community/LongCat-Flash-Lite-Sparse-4bit 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("mlx-community/LongCat-Flash-Lite-Sparse-4bit")

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 mlx-community/LongCat-Flash-Lite-Sparse-4bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mlx-community/LongCat-Flash-Lite-Sparse-4bit"

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": "mlx-community/LongCat-Flash-Lite-Sparse-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use mlx-community/LongCat-Flash-Lite-Sparse-4bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "mlx-community/LongCat-Flash-Lite-Sparse-4bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "mlx-community/LongCat-Flash-Lite-Sparse-4bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "mlx-community/LongCat-Flash-Lite-Sparse-4bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use mlx-community/LongCat-Flash-Lite-Sparse-4bit 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 "mlx-community/LongCat-Flash-Lite-Sparse-4bit"

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 mlx-community/LongCat-Flash-Lite-Sparse-4bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use mlx-community/LongCat-Flash-Lite-Sparse-4bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mlx-community/LongCat-Flash-Lite-Sparse-4bit"

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 "mlx-community/LongCat-Flash-Lite-Sparse-4bit" \
  --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"

AlazarM commited on 4 days ago

Commit

8a585ca

verified ·

1 Parent(s): ba24610

Upload README.md with huggingface_hub

Browse files

Files changed (1) hide show

README.md +25 -11

README.md CHANGED Viewed

@@ -20,6 +20,8 @@ pipeline_tag: text-generation
 To our knowledge this is the **first working implementation of LongCat-Flash-Lite-Sparse in any framework** — no upstream serving stack (mlx-lm, vLLM, SGLang, llama.cpp) supports the `oe_embed_*` variant yet.
 ## What's in this checkpoint
 LongCat-Flash-Lite-Sparse adds three things vanilla LongCat-Flash lacks:
 - **LongCat Sparse Attention (LSA)** — a DeepSeek-style lightning indexer over MLA, with streaming-aware indexing (fixed sink + local window) and cross-layer index reuse (`cli_factor`). Native long context.
@@ -48,20 +50,32 @@ print(generate(model, processor, text, max_tokens=64, temperature=0.0))
 # -> The capital of France is Paris.
 ```
-## Baseline (M5 Max, batch 1, greedy)
-Prefill (chunked at 512) and decode throughput vs. context length:
-| context | prefill tok/s | decode tok/s | peak mem |
 |--:|--:|--:|--:|
-| 512    | 1907 | 110.3 | 39.3 GB |
-| 1024   | 2724 |  99.0 | 39.4 GB |
-| 2048   | 1969 |  80.4 | 39.5 GB |
-| 4096   | 1781 |  82.5 | 39.8 GB |
-| 8192   | 1446 |  80.2 | 40.5 GB |
-| 16384  | 1067 |  76.7 | 41.8 GB |
-| 32768  |  571 |  72.6 | 44.2 GB |
-LSA's dynamic sparse selection activates once the KV length exceeds `index_topk` (2048), which keeps decode throughput nearly flat (110→73 tok/s) as context grows to 32k. Peak memory is the 4-bit weights (~36 GB) plus the MLA + indexer caches.
 ## License
 MIT, inherited from the base model.

 To our knowledge this is the **first working implementation of LongCat-Flash-Lite-Sparse in any framework** — no upstream serving stack (mlx-lm, vLLM, SGLang, llama.cpp) supports the `oe_embed_*` variant yet.
+4-bit (~36 GB of weights) is the smallest and fastest variant, for a 64 GB Mac. Also available: [6-bit](https://huggingface.co/AlazarM/LongCat-Flash-Lite-Sparse-6bit) (~52 GB, 96 GB Macs) and [8-bit](https://huggingface.co/AlazarM/LongCat-Flash-Lite-Sparse-8bit) (~68 GB, 128 GB Macs, near-lossless).
 ## What's in this checkpoint
 LongCat-Flash-Lite-Sparse adds three things vanilla LongCat-Flash lacks:
 - **LongCat Sparse Attention (LSA)** — a DeepSeek-style lightning indexer over MLA, with streaming-aware indexing (fixed sink + local window) and cross-layer index reuse (`cli_factor`). Native long context.
 # -> The capital of France is Paris.
 ```
+## Throughput (M5 Max, 128 GB, batch 1, greedy)
+Same methodology across quantizations (chunk-512 prefill, warmed kernels).
+**Decode tok/s**
+| ctx | 4-bit | 6-bit | 8-bit |
 |--:|--:|--:|--:|
+| 512   | 112 | 87 | 80 |
+| 1024  | 101 | 83 | 75 |
+| 2048  |  85 | 72 | 65 |
+| 4096  |  84 | 72 | 65 |
+| 8192  |  83 | 71 | 65 |
+| 16384 |  79 | 66 | 64 |
+| 32768 |  73 | 64 | 60 |
+**Prefill tok/s**
+| ctx | 4-bit | 6-bit | 8-bit |
+|--:|--:|--:|--:|
+| 512   | 3142 | 2581 | 2421 |
+| 2048  | 2386 | 2366 | 1923 |
+| 8192  | 1756 | 1627 | 1312 |
+| 32768 |  623 |  504 |  492 |
+**Footprint** — peak memory across 512→32k: 4-bit ~39–45 GB · 6-bit ~56–63 GB · 8-bit ~74–80 GB.
+LSA's dynamic sparse selection activates once the KV length exceeds `index_topk` (2048), keeping decode nearly flat (4-bit 112→73 tok/s to 32k). Batch-1 decode is partly weight-bandwidth-bound, so lower precision is faster; higher precision trades that for quality — only ~3B params are active per token, so quant error has little room to hide.
 ## License
 MIT, inherited from the base model.