diff options
| author | Laurent Mazare <laurent.mazare@gmail.com> | 2024-04-24 20:54:24 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-04-24 20:54:24 +0200 |
| commit | cfab6e761696c18b1ce5d3a339ab57ef191ca749 (patch) | |
| tree | 2c12c2092807e6170adfe934d48591257a456cec | |
| parent | 11d4a3c588e9ecb5bcc5a57773973b48036155f0 (diff) | |
| download | candle-cfab6e761696c18b1ce5d3a339ab57ef191ca749.tar.gz candle-cfab6e761696c18b1ce5d3a339ab57ef191ca749.tar.bz2 candle-cfab6e761696c18b1ce5d3a339ab57ef191ca749.zip | |
Mention phi-v3 in the readmes. (#2122)
| -rw-r--r-- | README.md | 5 | ||||
| -rw-r--r-- | candle-examples/examples/phi/README.md | 7 |
2 files changed, 7 insertions, 5 deletions
@@ -66,7 +66,8 @@ We also provide a some command line based examples using state of the art models - [Gemma](./candle-examples/examples/gemma/): 2b and 7b general LLMs from Google Deepmind. - [RecurrentGemma](./candle-examples/examples/recurrent-gemma/): 2b and 7b Griffin based models from Google that mix attention with a RNN like state. -- [Phi-1, Phi-1.5, and Phi-2](./candle-examples/examples/phi/): 1.3b and 2.7b general LLMs with performance on par with LLaMA-v2 7b. +- [Phi-1, Phi-1.5, Phi-2, and Phi-3](./candle-examples/examples/phi/): 1.3b, + 2.7b, and 3.8b general LLMs with performance on par with 7b models. - [StableLM-3B-4E1T](./candle-examples/examples/stable-lm/): a 3b general LLM pre-trained on 1T tokens of English and code datasets. Also supports StableLM-2, a 1.6b LLM trained on 2T tokens, as well as the code variants. @@ -203,7 +204,7 @@ If you have an addition to this list, please submit a pull request. - LLaMA v1, v2, and v3 with variants such as SOLAR-10.7B. - Falcon. - StarCoder, StarCoder2. - - Phi 1, 1.5, and 2. + - Phi 1, 1.5, 2, and 3. - Mamba, Minimal Mamba - Gemma 2b and 7b. - Mistral 7b v0.1. diff --git a/candle-examples/examples/phi/README.md b/candle-examples/examples/phi/README.md index 70af6650..e7fd8e9e 100644 --- a/candle-examples/examples/phi/README.md +++ b/candle-examples/examples/phi/README.md @@ -1,8 +1,9 @@ # candle-phi: 1.3b and 2.7b LLM with state of the art performance for <10b models. -[Phi-1.5](https://huggingface.co/microsoft/phi-1_5) and -[Phi-2](https://huggingface.co/microsoft/phi-2) are language models using -only 1.3 and 2.7 billion parameters but with state of the art performance compared to +[Phi-1.5](https://huggingface.co/microsoft/phi-1_5), +[Phi-2](https://huggingface.co/microsoft/phi-2), and +[Phi-3](https://huggingface.co/microsoft/Phi-3-mini-4k-instruct) are language models using +only 1.3, 2.7, and 3.8 billion parameters but with state of the art performance compared to models with up to 10 billion parameters. The candle implementation provides both the standard version as well as a |