| | --- |
| | title: README |
| | emoji: 🏆 |
| | colorFrom: pink |
| | colorTo: red |
| | sdk: static |
| | pinned: false |
| | --- |
| | |
| | <div align="center"> |
| | <a href="https://mobilint.com"> |
| | <img src="https://github.com/mobilint/mblt-model-zoo/blob/master/assets/Mobilint_Logo_Primary.png?raw=true" |
| | width="50%" |
| | alt="mobilint" /> |
| | </a> |
| | </div> |
| | |
| | # About Mobilint, Inc. |
| |
|
| | At Mobilint, we are at the forefront of addressing the challenges posed by the suboptimal performance of current processors. |
| | By harnessing the power of high-performance NPUs, we are pushing the boundaries of AI technology, transforming the once theoretical into the practical. |
| | This not only enhances the safety and convenience of individuals but also ignites the imagination of engineers. |
| |
|
| | We firmly believe in the transformative potential of AI, envisioning a future where it revolutionizes our lives in diverse and meaningful ways. |
| | Committed to excellence and innovation, Mobilint is dedicated to developing and deploying AI acceleration technologies that are shaping a better, more advanced world for everyone. |
| |
|
| | # Model List |
| |
|
| | The following tables summarize Transformers' models available with our inference package **mblt-model-zoo**. |
| | We provide the models that are quantized with our advanced quantization techniques. Performance metrics will be provided in the future. |
| |
|
| | ## Text Generation |
| |
|
| | | Model | Model ID | Link | Main | W8 | W4V8 | Note | |
| | | ----- | -------- | ---- | ---- | -- | ---- | ---- | |
| | | EXAONE-3.5-2.4B-Instruct | `mobilint/EXAONE-3.5-2.4B-Instruct` | [Link](https://huggingface.co/mobilint/EXAONE-3.5-2.4B-Instruct) | W4V8 | O | O | | |
| | | EXAONE-3.5-7.8B-Instruct | `mobilint/EXAONE-3.5-7.8B-Instruct` | [Link](https://huggingface.co/mobilint/EXAONE-3.5-7.8B-Instruct) | W4V8 | O | O | | |
| | | EXAONE-4.0-1.2B | `mobilint/EXAONE-4.0-1.2B` | [Link](https://huggingface.co/mobilint/EXAONE-4.0-1.2B) | W8 | O | X | | |
| | | EXAONE-Deep-2.4B | `mobilint/EXAONE-Deep-2.4B` | [Link](https://huggingface.co/mobilint/EXAONE-Deep-2.4B) | W8 | O | O | | |
| | | EXAONE-Deep-7.8B | `mobilint/EXAONE-Deep-7.8B` | [Link](https://huggingface.co/mobilint/EXAONE-Deep-7.8B) | W8 | O | O | | |
| | | HyperCLOVAX-SEED-Text-Instruct-1.5B | `mobilint/HyperCLOVAX-SEED-Text-Instruct-1.5B` | [Link](https://huggingface.co/mobilint/HyperCLOVAX-SEED-Text-Instruct-1.5B) | W4V8 | X | O | | |
| | | Llama-3.1-8B-Instruct | `mobilint/Llama-3.1-8B-Instruct` | [Link](https://huggingface.co/mobilint/Llama-3.1-8B-Instruct) | W4V8 | O | O | | |
| | | Llama-3.2-1B-Instruct | `mobilint/Llama-3.2-1B-Instruct` | [Link](https://huggingface.co/mobilint/Llama-3.2-1B-Instruct) | W8 | O | X | | |
| | | Llama-3.2-3B-Instruct | `mobilint/Llama-3.2-3B-Instruct` | [Link](https://huggingface.co/mobilint/Llama-3.2-3B-Instruct) | W4V8 | O | O | | |
| | | Qwen2.5-0.5B-Instruct | `mobilint/Qwen2.5-0.5B-Instruct` | [Link](https://huggingface.co/mobilint/Qwen2.5-0.5B-Instruct) | W8 | O | X | | |
| | | Qwen2.5-1.5B-Instruct | `mobilint/Qwen2.5-1.5B-Instruct` | [Link](https://huggingface.co/mobilint/Qwen2.5-1.5B-Instruct) | W8 | O | X | | |
| | | Qwen2.5-3B-Instruct | `mobilint/Qwen2.5-3B-Instruct` | [Link](https://huggingface.co/mobilint/Qwen2.5-3B-Instruct) | W4V8 | O | O | | |
| | | Qwen2.5-7B-Instruct | `mobilint/Qwen2.5-7B-Instruct` | [Link](https://huggingface.co/mobilint/Qwen2.5-7B-Instruct) | W4V8 | O | O | | |
| | | Qwen3-0.6B | `mobilint/Qwen3-0.6B` | [Link](https://huggingface.co/mobilint/Qwen3-0.6B) | W8 | O | X | | |
| | | Qwen3-1.7B | `mobilint/Qwen3-1.7B` | [Link](https://huggingface.co/mobilint/Qwen3-1.7B) | W8 | O | X | | |
| | | Qwen3-4B | `mobilint/Qwen3-4B` | [Link](https://huggingface.co/mobilint/Qwen3-4B) | W4V8 | O | O | | |
| | | Qwen3-8B | `mobilint/Qwen3-8B` | [Link](https://huggingface.co/mobilint/Qwen3-8B) | W4V8 | O | O | | |
| | | c4ai-command-r7b-12-2024 | `mobilint/c4ai-command-r7b-12-2024` | [Link](https://huggingface.co/mobilint/c4ai-command-r7b-12-2024) | W8 | X | X | | |
| |
|
| | ## Automatic Speech Recognition |
| |
|
| | | Model | Model ID | Link | Note | |
| | | ----- | -------- | ------ | ---- | |
| | | whisper-small | `mobilint/whisper-small` | [Link](https://huggingface.co/mobilint/whisper-small) | | |
| |
|
| | ## Image-Text-to-Text |
| |
|
| | | Model | Model ID | Link | Note | |
| | | ----- | -------- | ------ | ---- | |
| | | aya-vision-8b | `mobilint/aya-vision-8b` | [Link](https://huggingface.co/mobilint/aya-vision-8b) | | |
| | | Qwen2-VL-2B-Instruct | `mobilint/Qwen2-VL-2B-Instruct` | [Link](https://huggingface.co/mobilint/Qwen2-VL-2B-Instruct) | Only supports 1 image input with (224, 224) size. Image input will be resized automatically by our overridden preprocessor. | |
| |
|
| | ## Image to Text |
| |
|
| | | Model | Model ID | Link | Note | |
| | | ----- | -------- | ------ | ---- | |
| | | blip-image-captioning-large | `mobilint/blip-image-captioning-large` | [Link](https://huggingface.co/mobilint/blip-image-captioning-large) | | |
| |
|
| | ## Fill Mask |
| |
|
| | | Model | Model ID | Link | Note | |
| | | ----- | -------- | ------ | ---- | |
| | | bert-base-uncased | `mobilint/bert-base-uncased` | [Link](https://huggingface.co/mobilint/bert-base-uncased) | | |
| | | bert-kor-base | `mobilint/bert-kor-base` | [Link](https://huggingface.co/mobilint/bert-kor-base) | | |
