scut-language-coreml

Qwen2.5-1.5B-Instruct (Qwen/Qwen2.5-1.5B-Instruct @ 989aa79, © 2024 Alibaba Cloud, Apache-2.0) converted to a stateful Core ML program by Scut's tools/slm/convert.py: int4 weights per block of 32, a 4,096-token context, fp16 compute. The weights are the original model's apart from the quantization.

It is the small language model behind Better language in the Scut iOS app, which downloads these files when the person turns the option on. The folder holds the compiled model (ScutLanguage.mlmodelc) and the tokenizer.

Files

Path	Bytes	SHA-256
`ScutLanguage.mlmodelc/analytics/coremldata.bin`	243	`29e143cf955a4f8848de2e8a3476407e6b61dec0d4ae8e6fa48b62b74c0f7cb4`
`ScutLanguage.mlmodelc/coremldata.bin`	745	`779423a5d6ae8ca73b62354a3791038302a39b81d400f07b0471d750ae7bbf23`
`ScutLanguage.mlmodelc/metadata.json`	3,812	`7268a736f8c742e0b535999a111008eb39bc248a812ad0364a6cf239a2533a34`
`ScutLanguage.mlmodelc/model.mil`	885,305	`bcd811af1fe993454fe3a6c1b6c4b81ed3d94db5ec2248463ff25ba194ec726e`
`ScutLanguage.mlmodelc/weights/weight.bin`	869,497,024	`13b9462549b5b80c23d6c98899ac6005a5f8526801db0d7b8b32c87d895eecc1`
`tokenizer.json`	7,031,645	`c0382117ea329cdf097041132f6d735924b697924d6f6fc3945713e96ce87539`

Total 877,418,774 bytes.

Licence

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for lkibbalam/scut-language-coreml

Base model

Qwen/Qwen2.5-1.5B

Finetuned

Qwen/Qwen2.5-1.5B-Instruct

Finetuned

(1935)

this model