{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":52818766,"defaultBranch":"master","name":"MSA","ownerLogin":"renan-campos","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2016-02-29T19:42:15.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/6934052?v=4","public":true,"private":false,"isOrgOwned":false},"refInfo":{"name":"","listCacheKey":"v0:1679780968.0","currentOid":""},"activityList":{"items":[{"before":"f95dc4608e127cb1b2efa16432111cdf5cd8a4d4","after":"9110ccad291a9dc4de59e1bf23de812de68988df","ref":"refs/heads/master","pushedAt":"2023-08-01T17:24:16.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"renan-campos","name":"Renan Campos","path":"/renan-campos","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/6934052?s=80&v=4"},"commit":{"message":"Made the review_to_sentiment script more robust.\n\nDerived sentiment values for all reviews.","shortMessageHtmlLink":"Made the review_to_sentiment script more robust."}},{"before":"528a9c44e530914cb353bc1aae6b3ec27fe3d774","after":"f95dc4608e127cb1b2efa16432111cdf5cd8a4d4","ref":"refs/heads/master","pushedAt":"2023-04-03T12:52:29.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"renan-campos","name":"Renan Campos","path":"/renan-campos","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/6934052?s=80&v=4"},"commit":{"message":"Made the review_to_sentiment script more robust.\n\nDerived sentiment values for all reviews.","shortMessageHtmlLink":"Made the review_to_sentiment script more robust."}},{"before":"21aa2fdcdee52747fc074e91bc8839530f26945c","after":"528a9c44e530914cb353bc1aae6b3ec27fe3d774","ref":"refs/heads/master","pushedAt":"2023-03-28T20:33:37.463Z","pushType":"push","commitsCount":1,"pusher":{"login":"renan-campos","name":"Renan Campos","path":"/renan-campos","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/6934052?s=80&v=4"},"commit":{"message":"First draft of a script to label the sentiment test data using chat-gpt.\n\nThe script to check the token cost totaled 3,272,563 tokens. At $0.002\nper 1,000 tokens, the total cost to label all of the test data will be\n~$6.55. My openai account is granted $18 until June 1, so this will be\ncovered!\n\nMy initial script made it to 33 entries before it hit a rate limit\nerror. Only 20 requests can be sent per minute. With the help of chat\ngpt, I wrote some exponential backoff logic to get around this issue.\n\nI kicked off the script again and went snowboarding. It's surreal: Here\nI am on a mountain, and back at the cabin my laptop is interfacing with\none of the most advanced pieces of technology to date... to say if a\nmovie review was positive or negative.\n\nMy script needs improvements. I'd like to know how long the whole job\nwill have taken. It might be a good idea to save the raw responses too.\n\nWhen I got back, the script had exited at 388. My account says $0.26 has\nbeen spent so far. 117,822 tokens ~ $0.2356. The other $0.03 was\nprobably from the initial 33 entries evaluated.\n\nThe script exited on a gateway error. My script does what it needs to,\nbut a more robust design is needed.\n\nSkills are being built here: How to prompt an AI to perform the task you\nneed it to, and how to design a script that is robust to errors.","shortMessageHtmlLink":"First draft of a script to label the sentiment test data using chat-gpt."}},{"before":"bacbc7902ff569e9b125b0125fb6778b22661230","after":"21aa2fdcdee52747fc074e91bc8839530f26945c","ref":"refs/heads/master","pushedAt":"2023-03-27T14:49:53.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"renan-campos","name":"Renan Campos","path":"/renan-campos","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/6934052?s=80&v=4"},"commit":{"message":"Added script to calculate the token cost of sending to ChatGPT.\n\nI don't have an answer key for the test data set in this repository.\nWithout the answer key, models I create can't be evaluated with this\ndataset. I could manually read the texts and label them, but there are\n11,000 files.\n\nI see this as a good opportunity to learn and use the chat-gpt api. Is\nthis within the terms and conditions? ...I asked GPT and got no\ncomplaints!\n\nTo use the API, a model must be specified. GPT-4 is in limited beta, so\nI'll use gpt-3.5-turbo. This model may change, so it will be better to\nuse a tagged version, such as gpt-3.5-turbo-0301.\n\nAPI calls incur a cost. The cost is counted in tokens sent and received.\nA single call can have a maximum of 4096 tokens. My account is granted\n$18 until June 1 2023. It doesn't look like anything has been spent so\nfar!\n\nThis presents a potential problem: could the movie reviews I want to\nsend it be too long?\n\nThe script in this commit checks the cost of each review. This is\nwritten to a csv file for me to process later. A cursory look at the csv\nshows that none of the files ever breach 2000 tokens, so the API calls\nwill be safe!\n\nThe script was made with the help of Chat GPT. It's been a while since\nI've written python code. I've forgotten how to do many common\noperations, but I know what can be done. GPT has been a great help in\nthis regard. I learned a lot, and even corrected a provided example. The\noverall experience was highly engaging. I imagine this is what peer\nprogramming is like, but interacting with GPT has the advantage of no\nsocial pressue.\n\nI stay organized with the GPT conversations:\n  [vim help]\n  [nm-cli help]\n  [python help]\nThis helps maintain the context of the conversation to a certain topic.\nNot sure if it makes a difference on the GPT side, but it can't hurt...\nplus it helps me reference past interactions.","shortMessageHtmlLink":"Added script to calculate the token cost of sending to ChatGPT."}}],"hasNextPage":false,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAADYbQU4AA","startCursor":null,"endCursor":null}},"title":"Activity · renan-campos/MSA"}