{"items":[{"id":"e38907ad-d4a3-496f-b2fc-a1d2785b52d5","type":"push","org":"zhengqunkoo","repo":"tweet-prediction","title":"Fix beam search logic bug in metadata processing","summary":"Fixed a critical logic error in `beam_search` where the `new_top_k` dictionary was being prematurely reset inside the iteration loop. By moving the initialization outside the loop, the algorithm now correctly aggregates all candidates before performing the top-k selection. This ensures the beam search maintains coherence across candidates throughout the expansion process. ![Debugging](https://programmerhumor.io/wp-content/uploads/2024/05/programmerhumor-io-debugging-memes-testing-memes-d733cf88ed07de6.jpg)","url":"https://nomit.dev/zhengqunkoo/tweet-prediction/status/a0fe7d67c3a866e142a1d57d4c72b8d02b437aba4da3f27dec1febec6459905f","author":"zhengqunkoo","contributors":["zhengqunkoo"],"updated_at":"2017-06-09T14:06:28+00:00"},{"id":"4df5963c-7c1b-4cec-beb9-e6909c935673","type":"push","org":"zhengqunkoo","repo":"tweet-prediction","title":"Updated submission configuration parameters in gcloud batch script","summary":"The `gcloud/bash` script has been updated to reflect changes in the `make_submissions.py` interface. The command now uses the 'test' target and updates the k and j parameters to 10 respectively, ensuring better alignment with current processing requirements for submission generation.","url":"https://nomit.dev/zhengqunkoo/tweet-prediction/status/25659f04f74234bde6a7c7390218894b598b1a6415b83e7ce38a2b1d16ba28d6","author":"zhengqunkoo","contributors":["zhengqunkoo"],"updated_at":"2017-06-09T09:09:30+00:00"},{"id":"d1cc266b-4491-4fed-834c-09b62394eb98","type":"push","org":"zhengqunkoo","repo":"tweet-prediction","title":"Update to prediction handling and submission processing for model evaluation","summary":"Improved the flexibility of model evaluation by allowing beam search parameter 'k' to exceed 3 while still restricting output to the top 3 predictions. Also updated 'make_submissions' to support batch processing multiple .pickle files simultaneously, streamlining the submission workflow for large datasets.","url":"https://nomit.dev/zhengqunkoo/tweet-prediction/status/811d54fe21475e779ba6b24ecade2da8354a407917a768f9d237e45b159e7e7e","author":"zhengqunkoo","contributors":["zhengqunkoo"],"updated_at":"2017-06-09T08:55:08+00:00"},{"id":"709d0cdf-69b1-496d-889c-86560e8a8538","type":"push","org":"zhengqunkoo","repo":"tweet-prediction","title":"Refactored beam search output parsing with new strip_prediction utility","summary":"Updated the parsing logic in gcloud/metadata_preproc.py to use a more robust `strip_prediction` function for handling sequence delimiters. This change streamlines how model outputs are processed during beam search, replacing brittle indexing with a dedicated utility. It ensures cleaner output handling and resolves previous issues with trailing characters in predictions. ![Hero](https://i.ytimg.com/vi/iKwbWfN3dyY/oar2.jpg?sqp=-oaymwEkCJUDENAFSFqQAgHyq4qpAxMIARUAAAAAJQAAyEI9AICiQ3gB&rs=AOn4CLBg2EihRN7W7QV3HSJ2bRg3nJYhD3gB)","url":"https://nomit.dev/zhengqunkoo/tweet-prediction/status/d0da62273f490249c28f59536d2ee502c5d0c91ec45b08f3faa3f07a5d9c209d","author":"zhengqunkoo","contributors":["zhengqunkoo"],"updated_at":"2017-06-09T08:21:57+00:00"},{"id":"7f888beb-e5d3-4809-8909-236223e733f9","type":"push","org":"zhengqunkoo","repo":"tweet-prediction","title":"Refined output parsing to exclude empty or non-content word tokens","summary":"Updated the `parse_output` function in `metadata_preproc.py` to ensure that empty strings and single-space characters are not captured as valid words during post-processing. This prevents downstream issues where noise or whitespace might be incorrectly interpreted as content tokens.","url":"https://nomit.dev/zhengqunkoo/tweet-prediction/status/0808cfc350d1891952fa1b77d2f4067f943b07fd45b9cd27764f6675e91f46a4","author":"zhengqunkoo","contributors":["zhengqunkoo"],"updated_at":"2017-06-09T07:28:35+00:00"}],"pagination":{"offset":0,"limit":5,"has_more":true}}