Instructions to use 0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6 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("0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6")

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 0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6 with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use 0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6 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 "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6"

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 0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6

Run Hermes

hermes

OpenClaw new

How to use 0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6 with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6"

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 "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6" \
  --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"

MLX LM

How to use 0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6 with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "0xdfi/Qwen3.5-35B-A3B-REAP-pile10k-30p-MLX-q6",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

0xdfi commited on Mar 12

Commit

d729283

verified ·

1 Parent(s): 21c2c91

Refine wording in model card

Browse files

Files changed (1) hide show

README.md +3 -3

README.md CHANGED Viewed

@@ -54,7 +54,7 @@ The workflow for this release was:
 4. Apply that pruning plan to physically remove MoE experts from the checkpoint.
 5. Quantize the resulting pruned bf16 checkpoint into `q6`.
-This is **static MoE expert pruning**. It is **not** a runtime cartridge system, not hot-swapped experts, and not vLLM sleep-mode multiplexing.
 ## Calibration Data and How It Was Used
@@ -87,8 +87,8 @@ where `g_j(x)` is the router weight for expert `j` and `f_j(x)` is the expert ou
 ## Benchmark / Evaluation Status
-- Donna benchmark (held-out pile-10k + full HellaSwag): perplexity `9.441`, token accuracy retention `96.92%` of original bf16, HellaSwag retention `97.56%` of original bf16, throughput `123.95%` of original bf16, memory `29.97%` of original bf16.
-- Long-context Codex test (4 deterministic 100k+ token coding tasks): average score `83.74`, which is `85.64%` of original q6 and clearly above the simple same-budget baseline (`37.50`).
 ## Notes

 4. Apply that pruning plan to physically remove MoE experts from the checkpoint.
 5. Quantize the resulting pruned bf16 checkpoint into `q6`.
+This is **static MoE expert pruning**.
 ## Calibration Data and How It Was Used
 ## Benchmark / Evaluation Status
+- Custom benchmark (held-out pile-10k + full HellaSwag): perplexity `9.441`, token accuracy retention `96.92%` of original bf16, HellaSwag retention `97.56%` of original bf16, throughput `123.95%` of original bf16, memory `29.97%` of original bf16.
+- Custom long-context coding test (4 deterministic 100k+ token coding tasks): average score `83.74`, which is `85.64%` of original q6 and clearly above the simple same-budget baseline (`37.50`).
 ## Notes