in-context-impersonation
/
environment.yaml
72 строки · 1.9 Кб
1# reasons you might want to use `environment.yaml` instead of `requirements.txt`:
2# - pip installs packages in a loop, without ensuring dependencies across all packages
3# are fulfilled simultaneously, but conda achieves proper dependency control across
4# all packages
5# - conda allows for installing packages without requiring certain compilers or
6# libraries to be available in the system, since it installs precompiled binaries
7
8name: characters119
10channels:11- nvidia12- pytorch13- conda-forge14- defaults15
16# it is strongly recommended to specify versions of packages installed through conda
17# to avoid situation when version-unspecified packages install their latest major
18# versions which can sometimes break things
19
20# current approach below keeps the dependencies in the same major versions across all
21# users, but allows for different minor and patch versions of packages where backwards
22# compatibility is usually guaranteed
23
24dependencies:25- mamba=1.*26- pytorch=2.*27- torchvision=0.*28- pytorch-cuda=11.8.*29- lightning=2.*30- torchmetrics=0.*31- hydra-core=1.*32- rich=13.*33- pre-commit=3.*34- pytest=7.*35- pytest-xdist=3.*36
37# --------- loggers --------- #38- wandb=0.*39# - neptune-client40# - mlflow41# - comet-ml42
43# --------- code health --------- #44- black=23.*45- icontract=2.*46- mypy=1.*47- rope=1.*48
49# --------- libraries --------- #50- openai=0.*51- pandas=2.*52- seaborn=0.*53- spacy=3.*54- transformers=4.*55- textstat==0.*56- tqdm=4.*57
58- pip>=2359- pip:60- bitsandbytes==0.*61- farrow-and-ball==0.*62- fschat==0.2.2363- google-cloud-aiplatform>=1.26.164- hydra-optuna-sweeper==1.*65- hydra-colorlog==1.*66- hydra-submitit-launcher==1.*67- git+https://github.com/Lightning-AI/lit-llama.git68- langchain==0.*69- pyrootutils==1.*70- sentence_transformers==2.*71- slurm-gpustat==0.*72- gymnasium==0.*73