litert-community/Hammer2.1-1.5b

This model provides a few variants of MadeAgents/Hammer2.1-1.5b that are ready for deployment on Android using the LiteRT (fka TFLite) stack, MediaPipe LLM Inference API and LiteRT-LM.

Use the models

Colab

Disclaimer: The target deployment surface for the LiteRT models is Android/iOS/Web and the stack has been optimized for performance on these targets. Trying out the system in Colab is an easier way to familiarize yourself with the LiteRT stack, with the caveat that the performance (memory and latency) on Colab could be much worse than on a local device.

Android

Edge Gallery App

• Download or build the app from GitHub.

• Install the app from Google Play.

• Follow the instructions in the app.

LLM Inference API

• Download and install the apk.
• Follow the instructions in the app.

To build the demo app from source, please follow the instructions from the GitHub repository.

iOS

• Clone the MediaPipe samples repository and follow the instructions to build the LLM Inference iOS Sample App using XCode.
• Run the app via the iOS simulator or deploy to an iOS device.

Performance

Android

Note that all benchmark stats are from a Samsung S24 Ultra and multiple prefill signatures enabled.

Backend	Quantization scheme	Context length	Prefill (tokens/sec)	Decode (tokens/sec)	Time-to-first-token (sec)	Model size (MB)	Peak RSS Memory (MB)	GPU Memory (MB)
CPU	fp32 (baseline)	1280	51.50 tk/s	9.99 tk/s	20.30 s	6,180 MB	6252 MB	N/A	🔗
CPU	dynamic_int8	1280	290.00 tk/s	34.47 tk/s	3.79 s	1598 MB	1998 MB	N/A	🔗
CPU	dynamic_int8	4096	162.90 tk/s	23.66 tk/s	6.54 s	1598 MB	2215 MB	N/A	🔗
GPU	dynamic_int8	1280	1648.95 tk/s	30.20 tk/s	3.21 s	1598 MB	1814 MB	1505 MB	🔗
GPU	dynamic_int8	4096	920.04 tk/s	27.00 tk/s	4.17 s	1598 MB	1866 MB	1659 MB	🔗

• For the list of supported quantization schemes see supported-schemes. For these models, we are using prefill signature lengths of 32, 128, 512 and 1280.
• Model Size: measured by the size of the .tflite flatbuffer (serialization format for LiteRT models)
• Memory: indicator of peak RAM usage
• The inference on CPU is accelerated via the LiteRT XNNPACK delegate with 4 threads
• Benchmark is run with cache enabled and initialized. During the first run, the time to first token may differ.