|
| 1 | +<!--Copyright 2025 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 contain specific syntax for our doc-builder (similar to MDX) that may not be |
| 13 | +rendered properly in your Markdown viewer. |
| 14 | +
|
| 15 | +--> |
| 16 | + |
| 17 | +# GraniteMoeShared |
| 18 | + |
| 19 | +## Overview |
| 20 | + |
| 21 | + |
| 22 | +The GraniteMoe model was proposed in [Power Scheduler: A Batch Size and Token Number Agnostic Learning Rate Scheduler](https://arxiv.org/abs/2408.13359) by Yikang Shen, Matthew Stallone, Mayank Mishra, Gaoyuan Zhang, Shawn Tan, Aditya Prasad, Adriana Meza Soria, David D. Cox and Rameswar Panda. |
| 23 | + |
| 24 | +Additionally this class GraniteMoeSharedModel adds shared experts for Moe. |
| 25 | + |
| 26 | +```python |
| 27 | +import torch |
| 28 | +from transformers import AutoModelForCausalLM, AutoTokenizer |
| 29 | + |
| 30 | +model_path = "ibm-research/moe-7b-1b-active-shared-experts" |
| 31 | +tokenizer = AutoTokenizer.from_pretrained(model_path) |
| 32 | + |
| 33 | +# drop device_map if running on CPU |
| 34 | +model = AutoModelForCausalLM.from_pretrained(model_path, device_map="auto") |
| 35 | +model.eval() |
| 36 | + |
| 37 | +# change input text as desired |
| 38 | +prompt = "Write a code to find the maximum value in a list of numbers." |
| 39 | + |
| 40 | +# tokenize the text |
| 41 | +input_tokens = tokenizer(prompt, return_tensors="pt") |
| 42 | +# generate output tokens |
| 43 | +output = model.generate(**input_tokens, max_new_tokens=100) |
| 44 | +# decode output tokens into text |
| 45 | +output = tokenizer.batch_decode(output) |
| 46 | +# loop over the batch to print, in this example the batch size is 1 |
| 47 | +for i in output: |
| 48 | + print(i) |
| 49 | +``` |
| 50 | + |
| 51 | +This HF implementation is contributed by [Mayank Mishra](https://huggingface.co/mayank-mishra), [Shawn Tan](https://huggingface.co/shawntan) and [Sukriti Sharma](https://huggingface.co/SukritiSharma). |
| 52 | + |
| 53 | + |
| 54 | +## GraniteMoeSharedConfig |
| 55 | + |
| 56 | +[[autodoc]] GraniteMoeSharedConfig |
| 57 | + |
| 58 | +## GraniteMoeSharedModel |
| 59 | + |
| 60 | +[[autodoc]] GraniteMoeSharedModel |
| 61 | + - forward |
| 62 | + |
| 63 | +## GraniteMoeSharedForCausalLM |
| 64 | + |
| 65 | +[[autodoc]] GraniteMoeSharedForCausalLM |
| 66 | + - forward |
0 commit comments