If it works, create an alias in the terminal alias openai=~/.This is how you can quickly find where the openai program exists outside of the scripts/utils folder in ~/./Python3.X/./openai Before you delete it will list all of the locations that openai exists.If it isn’t there try triggering a delete in pip via pip uninstall openai.Find where the openai program is stored.openai is installed via pip3 show openai.you are in the correct venv (if applicable).If you are getting an error message in your terminal that says openai: command not found AND you have confirmed that (Note: the new alias would have automatically loaded the next time I opened the terminal).Ĭonfirmed that I had correctly added the alias.Īnd the output confirmed that I was good to go! Since then, I have had no issues. Just swap ~/.bashrc with what is applicable for you.īack in the terminal, I loaded my new settings into my current terminal. Note if you aren’t using Bash, below are the location of other common shells: To fix this permanently, I needed to edit my shell’s config file. I figured that a quick use of alias would solve my issue and I could all openai again from any directory. I missed being able to just use openai without any extra steps. As you can see in the comments attached to the issue on zsh-users GitHub, changing permissions and owners is an easy way to solve the problem. This was a big relief because I could finetune my model, However, I found navigating to the ~.\local\bin directory cumbersome. openai in the local bin folder it would execute. Getting openai To Run AgainĪfter poking around a bit more I found that if I ran. 18 Ive just installed ttf-mscorefonts on ubuntu 18.04 server (not desktop) by running the following command: sudo apt install ttf-mscorefonts-installer Everything seems to have installed correctly. The path was correctly modified, but I still got the same error of c ommand not found. I tried adding both the utils location and local bin location of the openai package to my terminal pathĪnd then checking that the path was correctly modified This proved to be very strange since I was fine tuning and testing models without a hitch for a few weeks now.Ī quick check with pip ( pip3 show openai) indicated that the package was still there so I figured that this was likely a path issue. Instead of changing the default shell to be zsh (the proper thing to do), bash is left as the default, the entirety of the old. Things have been going smoothly until I opened up my terminal today and tried to finetune a new model with the standard openai api fine_tunes.create -t -m. I fail to see the value of what is going on there. Option.I have been working on fine tuning GPT3 for a research project recently. Once you have that, go to your iTerm 2 and paste your path on "send text at start" So to do this, you need to go on your bash (where you have your commands working) and get the PATH variable from env (use this command to do that: env | grep PATH). Next, we are still missing the commands from bash in zsh. As you can see on the picture below, you need to select command option and paste path of zsh shell (to find the path, you can do which zsh).Īt this point you will have the zsh as your default shell ONLY for iTerm 2 and you will have bash as the global default shell on default mac terminal app. Then go to the profile tab and go down to command. If the command is still not working it can mean it is not installed, deleted or in the worst cases the system directory has been deleted or modified. On iTerm 2, go to preferences (or command ,). The command is variant, depending on the Linux distro you're using. To have both commands and zsh in iTerm 2 from bash, you need to do the following: But if it's a VPS, you can switch to the root account and install sudo. If you like me, you will have two terminals app, one is the default terminal with bash as the default shell and another iTerm 2 with zsh as its shell.
0 Comments
Leave a Reply. |