Comments (11)
Problem 1: The default yaml configuration is an absolute path, which can be changed to the directory where you store the weights, such as ./hugging_cache/llama-2-7b
, or directly use the online download method of meta-llama/ LLAMA-2-7B-HF
.
Problem 2: Since the original dataset is about 1K pieces, the above script takes about 1-2h to run. If you only want to verify the validity of the method, you can set ds_size
to 10 and edit only 10 samples.
See https://github.com/zjunlp/EasyEdit/blob/main/examples/run_zsre_llama2.py#L17
from easyedit.
Thank you so much for your response.
from easyedit.
@pengzju 请问假如我需要编辑llama2-7b模型,我需要从huggingface的llama2-7b仓库中下载哪些文件到./hugging_cache/llama-2-7b中呢?需要下载所有文件吗?谢谢!
from easyedit.
正常来讲,您只需要下载蓝框内的文件。包括模型config,tokenizer config,模型权重等
from easyedit.
感谢您的答复,谢谢!
from easyedit.
Hello,
While trying to run python run_zsre_llama2.py
--editing_method=MEMIT
--hparams_dir=../hparams/MEMIT/llama-7b
--data_dir=./data
I am getting the error:
python3.9/site-packages/datasets/builder.py", line 539, in _create_builder_config
raise ValueError(
ValueError: BuilderConfig '20200501.en' not found. Available: ['20220301.aa', '20220301.ab', '20220301.ace', '20220301.ady', '20220301.af', '20220301.ak', '20220301.als', '20220301.am', '20220301.an', '20220301.ang', '20220301.ar', '20220301.arc', '20220301.arz', '20220301.as', '20220301.ast', '20220301.atj', '20220301.av', '20220301.ay', '20220301.az', '20220301.azb', '20220301.ba', '20220301.bar', '20220301.bat-smg', '20220301.bcl', '20220301.be', '20220301.be-x-old', '20220301.bg', '20220301.bh', '20220301.bi', '20220301.bjn', '20220301.bm', '20220301.bn', '20220301.bo', '20220301.bpy', '20220301.br', '20220301.bs', '20220301.bug', '20220301.bxr', '20220301.ca', '20220301.cbk-zam', '20220301.cdo', '20220301.ce', '20220301.ceb', '20220301.ch', '20220301.cho', '20220301.chr', '20220301.chy', '20220301.ckb', '20220301.co', '20220301.cr', '20220301.crh', '20220301.cs', '20220301.csb', '20220301.cu', '20220301.cv', '20220301.cy', '20220301.da', '20220301.de', '20220301.din', '20220301.diq', '20220301.dsb', '20220301.dty', '20220301.dv', '20220301.dz', '20220301.ee', '20220301.el', '20220301.eml', '20220301.en', '20220301.eo', '20220301.es', '20220301.et', '20220301.eu', '20220301.ext', '20220301.fa', '20220301.ff', '20220301.fi', '20220301.fiu-vro', '20220301.fj', '20220301.fo', '20220301.fr', '20220301.frp', '20220301.frr', '20220301.fur', '20220301.fy', '20220301.ga', '20220301.gag', '20220301.gan', '20220301.gd', '20220301.gl', '20220301.glk', '20220301.gn', '20220301.gom', '20220301.gor', '20220301.got', '20220301.gu', '20220301.gv', '20220301.ha', '20220301.hak', '20220301.haw', '20220301.he', '20220301.hi', '20220301.hif', '20220301.ho', '20220301.hr', '20220301.hsb', '20220301.ht', '20220301.hu', '20220301.hy', '20220301.ia', '20220301.id', '20220301.ie', '20220301.ig', '20220301.ii', '20220301.ik', '20220301.ilo', '20220301.inh', '20220301.io', '20220301.is', '20220301.it', '20220301.iu', '20220301.ja', '20220301.jam', '20220301.jbo', '20220301.jv', '20220301.ka', '20220301.kaa', '20220301.kab', '20220301.kbd', '20220301.kbp', '20220301.kg', '20220301.ki', '20220301.kj', '20220301.kk', '20220301.kl', '20220301.km', '20220301.kn', '20220301.ko', '20220301.koi', '20220301.krc', '20220301.ks', '20220301.ksh', '20220301.ku', '20220301.kv', '20220301.kw', '20220301.ky', '20220301.la', '20220301.lad', '20220301.lb', '20220301.lbe', '20220301.lez', '20220301.lfn', '20220301.lg', '20220301.li', '20220301.lij', '20220301.lmo', '20220301.ln', '20220301.lo', '20220301.lrc', '20220301.lt', '20220301.ltg', '20220301.lv', '20220301.mai', '20220301.map-bms', '20220301.mdf', '20220301.mg', '20220301.mh', '20220301.mhr', '20220301.mi', '20220301.min', '20220301.mk', '20220301.ml', '20220301.mn', '20220301.mr', '20220301.mrj', '20220301.ms', '20220301.mt', '20220301.mus', '20220301.mwl', '20220301.my', '20220301.myv', '20220301.mzn', '20220301.na', '20220301.nah', '20220301.nap', '20220301.nds', '20220301.nds-nl', '20220301.ne', '20220301.new', '20220301.ng', '20220301.nl', '20220301.nn', '20220301.no', '20220301.nov', '20220301.nrm', '20220301.nso', '20220301.nv', '20220301.ny', '20220301.oc', '20220301.olo', '20220301.om', '20220301.or', '20220301.os', '20220301.pa', '20220301.pag', '20220301.pam', '20220301.pap', '20220301.pcd', '20220301.pdc', '20220301.pfl', '20220301.pi', '20220301.pih', '20220301.pl', '20220301.pms', '20220301.pnb', '20220301.pnt', '20220301.ps', '20220301.pt', '20220301.qu', '20220301.rm', '20220301.rmy', '20220301.rn', '20220301.ro', '20220301.roa-rup', '20220301.roa-tara', '20220301.ru', '20220301.rue', '20220301.rw', '20220301.sa', '20220301.sah', '20220301.sat', '20220301.sc', '20220301.scn', '20220301.sco', '20220301.sd', '20220301.se', '20220301.sg', '20220301.sh', '20220301.si', '20220301.simple', '20220301.sk', '20220301.sl', '20220301.sm', '20220301.sn', '20220301.so', '20220301.sq', '20220301.sr', '20220301.srn', '20220301.ss', '20220301.st', '20220301.stq', '20220301.su', '20220301.sv', '20220301.sw', '20220301.szl', '20220301.ta', '20220301.tcy', '20220301.te', '20220301.tet', '20220301.tg', '20220301.th', '20220301.ti', '20220301.tk', '20220301.tl', '20220301.tn', '20220301.to', '20220301.tpi', '20220301.tr', '20220301.ts', '20220301.tt', '20220301.tum', '20220301.tw', '20220301.ty', '20220301.tyv', '20220301.udm', '20220301.ug', '20220301.uk', '20220301.ur', '20220301.uz', '20220301.ve', '20220301.vec', '20220301.vep', '20220301.vi', '20220301.vls', '20220301.vo', '20220301.wa', '20220301.war', '20220301.wo', '20220301.wuu', '20220301.xal', '20220301.xh', '20220301.xmf', '20220301.yi', '20220301.yo', '20220301.za', '20220301.zea', '20220301.zh', '20220301.zh-classical', '20220301.zh-min-nan', '20220301.zh-yue', '20220301.zu']
Please suggest what should be done.
Thank you!
from easyedit.
Do you download the key status for layers [4, 5, 6, 7, 8] in llama2?
examples
├── data
│ ├── stats
│ │ └── ._hugging_cache_llama-2-7b
│ │ └── wikipedia_stats
│ │ ├── model.layers.4.mlp.down_proj_float32_mom2_100000.npz
│ │ ├── model.layers.5.mlp.down_proj_float32_mom2_100000.npz
│ │ ├── model.layers.6.mlp.down_proj_float32_mom2_100000.npz
│ │ ├── model.layers.7.mlp.down_proj_float32_mom2_100000.npz
│ │ └── model.layers.8.mlp.down_proj_float32_mom2_100000.npz
└── └── zsre_mend_eval_portability_gpt4.json
from easyedit.
![Screenshot 2023-08-26 at 12 35 32 AM](https://private-user-images.githubusercontent.com/53521802/263449466-1373278f-474f-4c80-b9fc-d6c9e645e9fa.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.GG7HYhPPTZMCNSCrvgEWESZvB7t7NrEMnwod6mcU-AU)
from easyedit.
Make sure that the directory for your load model in yaml is ./hugging_cache/llama-2-7b
, you can refer to the code here.
- I also recommend that you add a breakpoint here to see if
filename
exists. If it can route tofilename
, you do not need to download the wikipedia dataset, and your error will not occur.
from easyedit.
If your problem has been resolved, please help close this issue.
from easyedit.
Hi,
Yes, it has been resolved now. Thank you so much for the guidance and the quick response!
from easyedit.
Related Issues (20)
- locality and portability evaluation HOT 8
- GRACE sequential edit result HOT 15
- IKE fluency HOT 2
- ccks gpt2-xl 模型 为什么用roberta模型加载? HOT 2
- R-ROME has poor performance when using GPT2-xl HOT 2
- WISE CONTEXT_TEMPLATES_CACHE HOT 7
- WISE tokenize HOT 7
- what is the meaning of archive HOT 2
- [Wise] Editing loss = nan + 0 HOT 2
- AttributeError: Can't pickle local object 'length_collation.<locals>.collate_fn' HOT 8
- Issue with KN batch editing HOT 2
- Question about ZSRE experiment, how to reproduce results in your WISE paper? HOT 6
- PermissionError: [Errno 13] Permission denied: './results/models/MEND/gpt2-xl' HOT 3
- Using MEND with monkeypatch from higher HOT 10
- CCKS2024 ROME Question HOT 6
- WISE errors HOT 3
- Use EasyEdit with KnowEdit HOT 13
- Error calculation "adj_k" to edit LLAMA-3-8B using MEMIT HOT 4
- 关于wise.py HOT 3
- 一些关于KnowEditDataset类的疑问 HOT 3
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from easyedit.