|
| 1 | +<!--Copyright 2024 The HuggingFace Team. All rights reserved. |
| 2 | +
|
| 3 | +Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with |
| 4 | +the License. You may obtain a copy of the License at |
| 5 | +
|
| 6 | +http://www.apache.org/licenses/LICENSE-2.0 |
| 7 | +
|
| 8 | +Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on |
| 9 | +an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the |
| 10 | +specific language governing permissions and limitations under the License. |
| 11 | +
|
| 12 | +⚠️ Note that this file is in Markdown but contains specific syntax for our doc-builder (similar to MDX) that may not be |
| 13 | +rendered properly in your Markdown viewer. |
| 14 | +
|
| 15 | +--> |
| 16 | + |
| 17 | +# PhiMoE |
| 18 | + |
| 19 | +## Overview |
| 20 | + |
| 21 | +The PhiMoE model was proposed in [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://arxiv.org/abs/2404.14219) by Microsoft. |
| 22 | + |
| 23 | +### Summary |
| 24 | + |
| 25 | +The abstract from the Phi-3 paper is the following: |
| 26 | + |
| 27 | +We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide parameter-scaling results with a 7B, 14B models trained for 4.8T tokens, called phi-3-small, phi-3-medium, both significantly more capable than phi-3-mini (e.g., respectively 75%, 78% on MMLU, and 8.7, 8.9 on MT-bench). To enhance multilingual, multimodal, and long-context capabilities, we introduce three models in the phi-3.5 series: phi-3.5-mini, phi-3.5-MoE, and phi-3.5-Vision. The phi-3.5-MoE, a 16 x 3.8B MoE model with 6.6 billion active parameters, achieves superior performance in language reasoning, math, and code tasks compared to other open-source models of similar scale, such as Llama 3.1 and the Mixtral series, and on par with Gemini-1.5-Flash and GPT-4o-mini. Meanwhile, phi-3.5-Vision, a 4.2 billion parameter model derived from phi-3.5-mini, excels in reasoning tasks and is adept at handling both single-image and text prompts, as well as multi-image and text prompts. |
| 28 | + |
| 29 | +The original code for PhiMoE can be found [here](https://huggingface.co/microsoft/Phi-3.5-MoE-instruct). |
| 30 | + |
| 31 | +## Usage tips |
| 32 | + |
| 33 | +- This model is very similar to `Mixtral` with the main difference of [`Phi3LongRoPEScaledRotaryEmbedding`], where they are used to extend the context of the rotary embeddings. The query, key and values are fused, and the MLP's up and gate projection layers are also fused. |
| 34 | +- The tokenizer used for this model is identical to the [`LlamaTokenizer`], with the exception of additional tokens. |
| 35 | + |
| 36 | +## How to use PhiMoE |
| 37 | + |
| 38 | +<Tip warning={true}> |
| 39 | + |
| 40 | +Phi-3.5-MoE-instruct has been integrated in the development version (4.44.2.dev) of `transformers`. Until the official version is released through `pip`, ensure that you are doing the following: |
| 41 | +* When loading the model, ensure that `trust_remote_code=True` is passed as an argument of the `from_pretrained()` function. |
| 42 | + |
| 43 | +The current `transformers` version can be verified with: `pip list | grep transformers`. |
| 44 | + |
| 45 | +Examples of required packages: |
| 46 | +``` |
| 47 | +flash_attn==2.5.8 |
| 48 | +torch==2.3.1 |
| 49 | +accelerate==0.31.0 |
| 50 | +transformers==4.43.0 |
| 51 | +``` |
| 52 | + |
| 53 | +</Tip> |
| 54 | + |
| 55 | +```python |
| 56 | +import torch |
| 57 | +from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline |
| 58 | + |
| 59 | +torch.random.manual_seed(0) |
| 60 | + |
| 61 | +model = AutoModelForCausalLM.from_pretrained( |
| 62 | + "microsoft/Phi-3.5-MoE-instruct", |
| 63 | + device_map="cuda", |
| 64 | + torch_dtype="auto", |
| 65 | + trust_remote_code=True, |
| 66 | +) |
| 67 | + |
| 68 | +tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-3.5-MoE-instruct") |
| 69 | + |
| 70 | +messages = [ |
| 71 | + {"role": "system", "content": "You are a helpful AI assistant."}, |
| 72 | + {"role": "user", "content": "Can you provide ways to eat combinations of bananas and dragonfruits?"}, |
| 73 | + {"role": "assistant", "content": "Sure! Here are some ways to eat bananas and dragonfruits together: 1. Banana and dragonfruit smoothie: Blend bananas and dragonfruits together with some milk and honey. 2. Banana and dragonfruit salad: Mix sliced bananas and dragonfruits together with some lemon juice and honey."}, |
| 74 | + {"role": "user", "content": "What about solving an 2x + 3 = 7 equation?"}, |
| 75 | +] |
| 76 | + |
| 77 | +pipe = pipeline( |
| 78 | + "text-generation", |
| 79 | + model=model, |
| 80 | + tokenizer=tokenizer, |
| 81 | +) |
| 82 | + |
| 83 | +generation_args = { |
| 84 | + "max_new_tokens": 500, |
| 85 | + "return_full_text": False, |
| 86 | + "temperature": 0.0, |
| 87 | + "do_sample": False, |
| 88 | +} |
| 89 | + |
| 90 | +output = pipe(messages, **generation_args) |
| 91 | +print(output[0]['generated_text']) |
| 92 | +``` |
| 93 | + |
| 94 | +## PhimoeConfig |
| 95 | + |
| 96 | +[[autodoc]] PhimoeConfig |
| 97 | + |
| 98 | +<frameworkcontent> |
| 99 | +<pt> |
| 100 | + |
| 101 | +## PhimoeModel |
| 102 | + |
| 103 | +[[autodoc]] PhimoeModel |
| 104 | + - forward |
| 105 | + |
| 106 | +## PhimoeForCausalLM |
| 107 | + |
| 108 | +[[autodoc]] PhimoeForCausalLM |
| 109 | + - forward |
| 110 | + - generate |
| 111 | + |
| 112 | +## PhimoeForSequenceClassification |
| 113 | + |
| 114 | +[[autodoc]] PhimoeForSequenceClassification |
| 115 | + - forward |
| 116 | + |
| 117 | +</pt> |
| 118 | +</frameworkcontent> |
0 commit comments