engine/docs/ISQ.md

In situ quantization

In situ quantization works by quantizing models inplace, with the chief benefit being reduced memory footprint when running the model. This enables larger model to be run on devices which would not fit the full weights, and may increase model inference performance.

An API is exposed on the Python and Rust APIs which provide the ability to dynamically re-ISQ models at runtime.

To set the ISQ type for individual layers, use a model topology.

Note: 🔥 AFQ (affine) quantization is designed to be fast on Metal but is only supported on Metal.

Automatic ISQ

Automatic ISQ is an opt-in feature that selects the most accurate and fastest quantization method for the platform.

If the provided ISQ value is a valid integer (one of 2, 3, 4, 5, 6, or 8), the best quantization type for the platform will be chosen. Note that the fallback is always a Q/K quantization. On Metal, for 2, 3, 4, 6, or 8 bits, fast AFQ is used.

cargo run --release --features ... -- -i --isq 4 plain -m meta-llama/Llama-3.2-3B-Instruct

ISQ quantization types

• AFQ2 (AFQ is only available on Metal)
• AFQ3
• AFQ4
• AFQ6
• AFQ8
• Q4_0
• Q4_1
• Q5_0
• Q5_1
• Q8_0
• Q8_1 (not available on CUDA)
• Q2K
• Q3K
• Q4K
• Q5K
• Q6K
• Q8K (not available on CUDA)
• HQQ4
• HQQ8
• FP8

cargo run --release --features ... -- -i --isq 4 plain -m meta-llama/Llama-3.2-3B-Instruct

When using ISQ, it will automatically load ISQ-able weights into CPU memory before applying ISQ. The ISQ application process moves the weights to device memory. This process is implemented to avoid memory spikes from loading the model in full precision.

For Mixture of Expert models, a method called MoQE can be applied to only quantize MoE layers. This is configured via the ISQ "organization" parameter in all APIs. The following models support MoQE:

• Phi 3.5 MoE
• V2
• V3 / R1

Accuracy

Accuracy of ISQ can be measured by the performance degradation versus the unquantized model. This is commonly measured with perplexity. Please see the perplexity example.

To improve the accuracy of a model with ISQ, use an imatrix file. These can be found online (for example, on Hugging Face), and should be passed with the --imatrix flag for plain models. This will increase the accuracy of the quantization significantly and bring the ISQ quantization up to par with the GGUF counterpart.

Check out the imatrix docs.

Python Example

runner = Runner(
    which=Which.Plain(
        model_id="Qwen/Qwen3-0.6B",
    ),
    in_situ_quant="4",
)

Rust Example

You can find this example here.

let model = TextModelBuilder::new("microsoft/Phi-3.5-mini-instruct")
    .with_isq(IsqType::Q8_0)
    .with_logging()
    .with_paged_attn(|| PagedAttentionMetaBuilder::default().build())?
    .build()
    .await?;

Server example

cargo run --release --features "cuda flash-attn" -- --port 1234 --log output.txt --isq Q2K plain -m mistralai/Mistral-7B-Instruct-v0.1