{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":618511002,"defaultBranch":"master","name":"dolly","ownerLogin":"databrickslabs","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2023-03-24T16:15:53.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/49501376?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1681758266.0","currentOid":""},"activityList":{"items":[{"before":"f9dc122f840e987b20aaf2ced6157043dd8b8675","after":"e2d664ddf06e0723d818dc56d6c522ead4bd881d","ref":"refs/heads/master","pushedAt":"2023-06-30T18:36:16.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Update README.md","shortMessageHtmlLink":"Update README.md"}},{"before":"b337d537471c67bd73d78217ed266766440be112","after":"f9dc122f840e987b20aaf2ced6157043dd8b8675","ref":"refs/heads/master","pushedAt":"2023-06-30T18:35:51.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Update README.md\n\nadd citation","shortMessageHtmlLink":"Update README.md"}},{"before":"aaa0ecb5a5555f99e57e6582f1fb3d289f31940f","after":"b337d537471c67bd73d78217ed266766440be112","ref":"refs/heads/master","pushedAt":"2023-06-24T03:01:22.000Z","pushType":"pr_merge","commitsCount":4,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #200 from edurdevic/update-readme-DBR-13\n\nUpdate README.md with instructions for DBR 13","shortMessageHtmlLink":"Merge pull request #200 from edurdevic/update-readme-DBR-13"}},{"before":"a35770657371743a7d691178036ce0befd342d60","after":"aaa0ecb5a5555f99e57e6582f1fb3d289f31940f","ref":"refs/heads/master","pushedAt":"2023-06-08T12:46:47.713Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #193 from srowen/DBR13\n\nRemove DBR 12 notes in favor of 13; remove test loading data as it doesn't respect flag anyway","shortMessageHtmlLink":"Merge pull request #193 from srowen/DBR13"}},{"before":"fd1a733a77ec5174dcb07d048315fde0c1ce6ea3","after":"a35770657371743a7d691178036ce0befd342d60","ref":"refs/heads/master","pushedAt":"2023-06-07T00:12:34.218Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #192 from srowen/bf16flag\n\nSet bf16 flags correctly for a10/a100","shortMessageHtmlLink":"Merge pull request #192 from srowen/bf16flag"}},{"before":"a33d774e857e1e19fcf74cf0357dc27c78a4cd06","after":"fd1a733a77ec5174dcb07d048315fde0c1ce6ea3","ref":"refs/heads/master","pushedAt":"2023-06-05T19:02:32.970Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #189 from holdenk/fix-requirements\n\nDrop duplicate torch from requirements_dev.txt","shortMessageHtmlLink":"Merge pull request #189 from holdenk/fix-requirements"}},{"before":"5021d941d95dddcf1f00d978d7f944709873f419","after":"a33d774e857e1e19fcf74cf0357dc27c78a4cd06","ref":"refs/heads/master","pushedAt":"2023-06-02T17:14:42.660Z","pushType":"pr_merge","commitsCount":23,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #182 from tnixon/a10_v100_config\n\nA10 & v100 config","shortMessageHtmlLink":"Merge pull request #182 from tnixon/a10_v100_config"}},{"before":"3725600f2c314237b1b3f611b7c42126be56b051","after":"5021d941d95dddcf1f00d978d7f944709873f419","ref":"refs/heads/master","pushedAt":"2023-05-26T16:40:58.713Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"[Fix] Move attention mask to the model device type (#180)\n\nThe attention mask needs to be on the same device as the rest of the model and inputs, or else there will be a device mismatch.","shortMessageHtmlLink":"[Fix] Move attention mask to the model device type (#180)"}},{"before":"eb8f146dfc8daa0077966fc1b89e0519c53c5124","after":"3725600f2c314237b1b3f611b7c42126be56b051","ref":"refs/heads/master","pushedAt":"2023-05-16T18:29:57.658Z","pushType":"pr_merge","commitsCount":4,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #164 from xuanyuanking/add_dolly_as_input_model\n\nAdd Dolly as the Input Model","shortMessageHtmlLink":"Merge pull request #164 from xuanyuanking/add_dolly_as_input_model"}},{"before":"fc22734003b5ef5731ccbb8ecb469d7aa2afe175","after":"eb8f146dfc8daa0077966fc1b89e0519c53c5124","ref":"refs/heads/master","pushedAt":"2023-05-15T14:24:11.610Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #162 from SamiKalliomaki/fix-docs\n\nFix documentation for max_new_tokens.","shortMessageHtmlLink":"Merge pull request #162 from SamiKalliomaki/fix-docs"}},{"before":"cad7b3fefe72b91cdc22f80e8c780ccb49881f92","after":"fc22734003b5ef5731ccbb8ecb469d7aa2afe175","ref":"refs/heads/master","pushedAt":"2023-05-03T21:09:24.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #144 from srowen/DatasetHF\n\nNote that dataset should be used from Hugging Face now","shortMessageHtmlLink":"Merge pull request #144 from srowen/DatasetHF"}},{"before":"5f9bfba300181fb35854d64c8bf1180ea75105f7","after":"cad7b3fefe72b91cdc22f80e8c780ccb49881f92","ref":"refs/heads/master","pushedAt":"2023-05-03T20:47:26.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Update to Trainer.train to Allow Override Dataset (#142)\n\nThis update adds a new parameter '--training-dataset' that can be optionally set at the deepspeed call to pass in an alternate dataset for training.\r\n\r\nAdds a new constant for the databricks 15k dataset. \r\n\r\nUpdates various Trainer functions to allow for a path override for the dataset.","shortMessageHtmlLink":"Update to Trainer.train to Allow Override Dataset (#142)"}},{"before":"34c0bd289ca4619a60983d3fdea1982f9b7dade6","after":"5f9bfba300181fb35854d64c8bf1180ea75105f7","ref":"refs/heads/master","pushedAt":"2023-04-25T16:28:21.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #130 from srowen/deepspeed083\n\nDrop back to deepspeed 0.8.3 because of issues with 0.9.x","shortMessageHtmlLink":"Merge pull request #130 from srowen/deepspeed083"}},{"before":"3fd1286c3a922cbf96b66fb9914ee7e75d477369","after":"34c0bd289ca4619a60983d3fdea1982f9b7dade6","ref":"refs/heads/master","pushedAt":"2023-04-21T23:31:02.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #123 from srowen/Dataset15k\n\nReference HF dataset by default, now that it's live","shortMessageHtmlLink":"Merge pull request #123 from srowen/Dataset15k"}},{"before":"56e7c0f40fd49d25572d7773c6035a8d652b38fd","after":"3fd1286c3a922cbf96b66fb9914ee7e75d477369","ref":"refs/heads/master","pushedAt":"2023-04-21T23:30:33.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #122 from srowen/UpdateReqs\n\nUpdate reqs to match DBR 13; add torch","shortMessageHtmlLink":"Merge pull request #122 from srowen/UpdateReqs"}},{"before":"d69a93b90e589cb958ea2de7781926d5a8ebd93e","after":"56e7c0f40fd49d25572d7773c6035a8d652b38fd","ref":"refs/heads/master","pushedAt":"2023-04-20T17:26:57.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #113 from srowen/FixPythia28\n\nFix reference to pythia-2.8b","shortMessageHtmlLink":"Merge pull request #113 from srowen/FixPythia28"}},{"before":"0eadcb7b0648d496d67243a7d572b413560be661","after":"d69a93b90e589cb958ea2de7781926d5a8ebd93e","ref":"refs/heads/master","pushedAt":"2023-04-19T19:02:46.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"srowen","name":"Sean Owen","path":"/srowen","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/822522?s=80&v=4"},"commit":{"message":"Merge pull request #106 from srowen/UpdateOtherTraining\n\nImprove batch size guidance for other instance training","shortMessageHtmlLink":"Merge pull request #106 from srowen/UpdateOtherTraining"}},{"before":"46d302849490b59e23f94e7a0e51ed1810a9f39f","after":null,"ref":"refs/heads/fix_get_special_token_id_exc","pushedAt":"2023-04-17T19:04:26.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"}},{"before":"dcef926686b30a652b4c93c27ce6e4cd7864bb01","after":"0eadcb7b0648d496d67243a7d572b413560be661","ref":"refs/heads/master","pushedAt":"2023-04-17T19:04:23.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"load_model_tokenizer_for_generate should raise ValueError (#99)\n\nThis is what InstructionTextGenerationPipeline expects.\r\n\r\nAlso same code used in: https://huggingface.co/databricks/dolly-v2-12b/blob/main/instruct_pipeline.py","shortMessageHtmlLink":"load_model_tokenizer_for_generate should raise ValueError (#99)"}},{"before":null,"after":"46d302849490b59e23f94e7a0e51ed1810a9f39f","ref":"refs/heads/fix_get_special_token_id_exc","pushedAt":"2023-04-17T19:03:34.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"load_model_tokenizer_for_generate should raise ValueError\n\nThis is what InstructionTextGenerationPipeline expects.","shortMessageHtmlLink":"load_model_tokenizer_for_generate should raise ValueError"}},{"before":"3ecb8d9556e883497973484b9e3d3aba02964ea3","after":"dcef926686b30a652b4c93c27ce6e4cd7864bb01","ref":"refs/heads/master","pushedAt":"2023-04-17T04:58:00.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Suggest bfloat16 and add generation notes for A10, V100 (#95)\n\nI propose we explicitly show loading in bf16 over fp32 in the generation example and in the generation code. This helps avoid OOM for sure in many usages.\r\n\r\nI add additional notes on getting generation to work on A10, V100 GPUs with 8-bit.","shortMessageHtmlLink":"Suggest bfloat16 and add generation notes for A10, V100 (#95)"}},{"before":"662d181ce7d026d394905ebc71b0d094ccc35580","after":"3ecb8d9556e883497973484b9e3d3aba02964ea3","ref":"refs/heads/master","pushedAt":"2023-04-17T03:25:47.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Update notes for training on A10/V100 (#94)\n\nThis is a quick pass as simplistic tweaks needed to train the smaller Dolly sizes on A10 or V100 instance types","shortMessageHtmlLink":"Update notes for training on A10/V100 (#94)"}},{"before":"905e58a1ef9f5f22ba39285855f39b20b41003fe","after":"662d181ce7d026d394905ebc71b0d094ccc35580","ref":"refs/heads/master","pushedAt":"2023-04-17T02:52:00.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Add NVIDIA library installs for DBR 13.0 (#93)\n\nThis just also adds a commented-out cell with the right NVIDIA lib installs for Databricks 13.0 ML","shortMessageHtmlLink":"Add NVIDIA library installs for DBR 13.0 (#93)"}},{"before":"4c3419cd7ec646647c13cfe54c4270296fa91524","after":null,"ref":"refs/heads/v2_updates","pushedAt":"2023-04-15T23:49:24.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"}},{"before":"3ea242cb551151bb2bc1e971693831b334ab903d","after":"905e58a1ef9f5f22ba39285855f39b20b41003fe","ref":"refs/heads/master","pushedAt":"2023-04-15T23:49:21.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Dolly V2 Updates (#88)\n\nThis updates training to use the [`databricks-dolly-15k`](https://github.com/databrickslabs/dolly/tree/master/data) dataset. It also includes improvements to text generation and example notebooks.\r\n\r\nKey Changes:\r\n* The `train_dolly.py` notebook now uses Pythia models as the input models and fine tunes using the [`databricks-dolly-15k`](https://github.com/databrickslabs/dolly/tree/master/data) dataset.\r\n* Added `InstructionTextGenerationPipeline` for text generation. This is derived from the code in the model repo, [instruct_pipeline.py](https://huggingface.co/databricks/dolly-v2-12b/blob/main/instruct_pipeline.py). It has been improved so that it is compatible with the `TextGenerationPipeline` from the `transformers` library. Some code, such as that in `_forward`, was copied from that pipeline to help with compatibility. The biggest change relative to the current `instruct_pipeline.py` version is that it returns a list of dicts per instruction, rather than just a dict. It also now has a `return_full_text` option. Both of these contribute towards being usable with `langchain`.\r\n* `generate_response` is now a wrapper around `InstructionTextGenerationPipeline`, as the code was all moved there.\r\n* `trainer.py` now uses the local `databricks-dolly-15k.jsonl` dataset. A `text` column has been constructed from the instruction, context, and response.\r\n\r\nMinor Changes:\r\n* Added an `experiment_id` widget to help keep track of different models that are fine tuned.\r\n* Added more options to CLI for configuring training.\r\n\r\nAdditional Changes:\r\n* Added a `generation.py` example notebook that uses `generate_response` on a couple instructions.\r\n* Added a `langchain.py` example notebook that uses `HuggingFacePipeline ` from `langchain` and `InstructionTextGenerationPipeline ` to test instructions both with and without context.\r\n* Added a `pipeline.py` example notebook that uses `InstructionTextGenerationPipeline` to generate multiple samples per instruction.","shortMessageHtmlLink":"Dolly V2 Updates (#88)"}},{"before":"8825316e8a2d2a15b22a0126f4d6d8087e529815","after":"4c3419cd7ec646647c13cfe54c4270296fa91524","ref":"refs/heads/v2_updates","pushedAt":"2023-04-15T23:49:07.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"fix return_full_text issue","shortMessageHtmlLink":"fix return_full_text issue"}},{"before":null,"after":"8825316e8a2d2a15b22a0126f4d6d8087e529815","ref":"refs/heads/v2_updates","pushedAt":"2023-04-15T23:03:02.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Dolly V2 Updates","shortMessageHtmlLink":"Dolly V2 Updates"}},{"before":"7bdfa8557c48954587d9e5a26f1daddf9daa0e15","after":"3ea242cb551151bb2bc1e971693831b334ab903d","ref":"refs/heads/master","pushedAt":"2023-04-14T19:34:59.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"fix typo in trainer.py (#81)\n\nseq_lenth -> seq_length","shortMessageHtmlLink":"fix typo in trainer.py (#81)"}},{"before":"5d3d5cf84bb80337e93984e512afc3a59be1246e","after":"7bdfa8557c48954587d9e5a26f1daddf9daa0e15","ref":"refs/heads/master","pushedAt":"2023-04-13T20:29:00.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"matthayes","name":"Matthew Hayes","path":"/matthayes","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/212331?s=80&v=4"},"commit":{"message":"Fix new line in language / version (#70)","shortMessageHtmlLink":"Fix new line in language / version (#70)"}},{"before":"2305eb7f2f4b3beb2379f34c6addf335b46c4b43","after":null,"ref":"refs/heads/data-update-apr13","pushedAt":"2023-04-13T17:22:06.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"mike-conover-db","name":"Mike Conover","path":"/mike-conover-db","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/102997399?s=80&v=4"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAADTH87XwA","startCursor":null,"endCursor":null}},"title":"Activity ยท databrickslabs/dolly"}