{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":750409062,"defaultBranch":"main","name":"contrastors","ownerLogin":"nomic-ai","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2024-01-30T15:39:06.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/102670180?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1717596454.0","currentOid":""},"activityList":{"items":[{"before":"89a7af13a90cf0ac5d8a549ede6835361bcd8413","after":null,"ref":"refs/heads/dependabot/pip/requests-2.32.2","pushedAt":"2024-06-05T14:07:34.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"}},{"before":"5c2eff040222c17d172a5b91b2035cc53a9e8070","after":"56cfae8fb70b6c86208c42f9cefa927206090ab8","ref":"refs/heads/main","pushedAt":"2024-06-05T14:07:33.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Merge pull request #41 from nomic-ai/dependabot/pip/requests-2.32.2\n\nbuild(deps): bump requests from 2.31.0 to 2.32.2","shortMessageHtmlLink":"Merge pull request #41 from nomic-ai/dependabot/pip/requests-2.32.2"}},{"before":"dc7299e61d052ef18564f2dd5411441684f72f1e","after":null,"ref":"refs/heads/dependabot/pip/pillow-10.3.0","pushedAt":"2024-06-05T14:06:59.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"}},{"before":"a547553ada51c8d9aecc752eaccb0e218397e50b","after":"5c2eff040222c17d172a5b91b2035cc53a9e8070","ref":"refs/heads/main","pushedAt":"2024-06-05T14:06:58.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Merge pull request #42 from nomic-ai/dependabot/pip/pillow-10.3.0\n\nbuild(deps): bump pillow from 10.2.0 to 10.3.0","shortMessageHtmlLink":"Merge pull request #42 from nomic-ai/dependabot/pip/pillow-10.3.0"}},{"before":null,"after":"dc7299e61d052ef18564f2dd5411441684f72f1e","ref":"refs/heads/dependabot/pip/pillow-10.3.0","pushedAt":"2024-06-05T13:55:39.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"dependabot[bot]","name":null,"path":"/apps/dependabot","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/29110?s=80&v=4"},"commit":{"message":"build(deps): bump pillow from 10.2.0 to 10.3.0\n\nBumps [pillow](https://github.com/python-pillow/Pillow) from 10.2.0 to 10.3.0.\n- [Release notes](https://github.com/python-pillow/Pillow/releases)\n- [Changelog](https://github.com/python-pillow/Pillow/blob/main/CHANGES.rst)\n- [Commits](https://github.com/python-pillow/Pillow/compare/10.2.0...10.3.0)\n\n---\nupdated-dependencies:\n- dependency-name: pillow\n dependency-type: direct:production\n...\n\nSigned-off-by: dependabot[bot] ","shortMessageHtmlLink":"build(deps): bump pillow from 10.2.0 to 10.3.0"}},{"before":null,"after":"89a7af13a90cf0ac5d8a549ede6835361bcd8413","ref":"refs/heads/dependabot/pip/requests-2.32.2","pushedAt":"2024-06-05T13:55:37.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"dependabot[bot]","name":null,"path":"/apps/dependabot","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/29110?s=80&v=4"},"commit":{"message":"build(deps): bump requests from 2.31.0 to 2.32.2\n\nBumps [requests](https://github.com/psf/requests) from 2.31.0 to 2.32.2.\n- [Release notes](https://github.com/psf/requests/releases)\n- [Changelog](https://github.com/psf/requests/blob/main/HISTORY.md)\n- [Commits](https://github.com/psf/requests/compare/v2.31.0...v2.32.2)\n\n---\nupdated-dependencies:\n- dependency-name: requests\n dependency-type: direct:production\n...\n\nSigned-off-by: dependabot[bot] ","shortMessageHtmlLink":"build(deps): bump requests from 2.31.0 to 2.32.2"}},{"before":"c545be2bc0074ea343eaa5247706965e13c1b5fa","after":"a547553ada51c8d9aecc752eaccb0e218397e50b","ref":"refs/heads/main","pushedAt":"2024-06-05T13:55:00.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"feat: image-text (#3)\n\n* fix: loading of in21k vit\r\n\r\n* fix: config for gh200\r\n\r\n* fix: hostfile\r\n\r\n* fix: hostfile\r\n\r\n* fix: requirements\r\n\r\n* fix: multinode\r\n\r\n* fix: impor tos\r\n\r\n* fix: model name arg, dataset size, text query\r\n\r\n* chore: ignore ids to keep\r\n\r\n* feat: build index and save embeddings + rankings\r\n\r\n* fix: update for cp medium\r\n\r\n* fix: update build_map_index_filter\r\n\r\n* fix: save faster, remove expensive gather\r\n\r\n* fix: option to save\r\n\r\n* fix: reqs\r\n\r\n* fix: protobuf\r\n\r\n* fix: beam req\r\n\r\n* fix: beam req\r\n\r\n* fix: faster multi-gpu index filteirng\r\n\r\n* feat: filter uuids from json files\r\n\r\n* fix: print statement\r\n\r\n* fix: remove duplicates\r\n\r\n* fix: remove unused\r\n\r\n* feat: precompute text embeddings\r\n\r\n* fix: remove unused\r\n\r\n* fix: updates\r\n\r\n* fix: download and process\r\n\r\n* fix: working now\r\n\r\n* feat: cross attention/coca-esque model\r\n\r\n* fix: reorder tar so npy near others\r\n\r\n* feat: add back sampling with replacement (need to test)\r\n\r\n* feat: train with pretrained\r\n\r\n* fix: precomputed\r\n\r\n* feat: cpu offload faster than regular opt w grad checkp\r\n\r\n* fix: can load and resample now!\r\n\r\n* feat: sugarcrepe eval\r\n\r\n* fix: prefix for imagenet\r\n\r\n* fix: add prefix in trainer for eval\r\n\r\n* feat: datacomp evals for contrators models\r\n\r\n* fix: prefix + print after evals\r\n\r\n* fix: if empty set to none\r\n\r\n* fix: dataset-size pass via cli\r\n\r\n* fix: fairness eval\r\n\r\n* fix: allow null path for imagenet for testing\r\n\r\n* feat: mlm + contrastive loss\r\n\r\n* fix: imagenet fixes\r\n\r\n* fix: deepspeed config\r\n\r\n* fix: imagenet eval\r\n\r\n* feat: three towers image <-> text, text <-> frozen\r\n\r\n* fix: eval steps\r\n\r\n* fix: hf model updates\r\n\r\n* fix: vit pos embed\r\n\r\n* feat: three towers current\r\n\r\n* fix: multinode fixes\r\n\r\n* fix: global rank in multinode\r\n\r\n* fix: progbar only global rank 0\r\n\r\n* feat: higher lr\r\n\r\n* fix: eval strategy epochs logging fix\r\n\r\n* feat: no clamp logits config\r\n\r\n* feat: 3 epoch training\r\n\r\n* feat: update hostfile\r\n\r\n* fix: 10 epochs\r\n\r\n* fix: update hostfile\r\n\r\n* feat: upload embs to atlas\r\n\r\n* feat: dino v1\r\n\r\n* fix: grad check\r\n\r\n* fix: clip model\r\n\r\n* feat: 32k vit-l\r\n\r\n* fix: update hostfile\r\n\r\n* fix: workers\r\n\r\n* fix: more logging\r\n\r\n* fix: no wandb for now\r\n\r\n* fix: try smaller vit\r\n\r\n* fix: try more ds stuff\r\n\r\n* fix: try openclip loss\r\n\r\n* fix: remove unneeded print\r\n\r\n* fix: test clip loss\r\n\r\n* fix: 32k run\r\n\r\n* fix: are evals broken?\r\n\r\n* fix: 16k testing\r\n\r\n* fix: evals\r\n\r\n* fix: evals\r\n\r\n* chore: logging\r\n\r\n* fix: remove prints\r\n\r\n* feat: config\r\n\r\n* fix: remove rng, trust openclip\r\n\r\n* fix: idk?\r\n\r\n* fix: path\r\n\r\n* fix: rank\r\n\r\n* feat: ok now working L14\r\n\r\n* feat: 32k higher lr exp\r\n\r\n* feat: fb vit mae\r\n\r\n* feat: mae train\r\n\r\n* fix: map mae\r\n\r\n* fix: sp\r\n\r\n* fix: batch size\r\n\r\n* feat: 10epoch 65k\r\n\r\n* feat: higher lr\r\n\r\n* feat: no wd\r\n\r\n* feat: long train\r\n\r\n* feat: 81k bs\r\n\r\n* feat: 3 epoch 65k\r\n\r\n* feat: 10 epoch\r\n\r\n* fix: large 3 epoch train\r\n\r\n* fix: workers\r\n\r\n* fix: model utils loading\r\n\r\n* fix: dataloader for datacomp1b\r\n\r\n* fix: remove pdb\r\n\r\n* fix: workers\r\n\r\n* feat: dfn 2b\r\n\r\n* fix: bs\r\n\r\n* fix: bs\r\n\r\n* fix: wandb\r\n\r\n* fix: imagenet workers\r\n\r\n* feat: try unidirectional\r\n\r\n* fix: path for old h100\r\n\r\n* fix: map\r\n\r\n* fix: lets try this again\r\n\r\n* fix: try fusing\r\n\r\n* fix: bad code\r\n\r\n* fix: 32k map fix\r\n\r\n* fix: bs and default get for dataset\r\n\r\n* fix: fused\r\n\r\n* fix; dumb\r\n\r\n* fix: try this\r\n\r\n* feat: pos embed with swiglu gated\r\n\r\n* fix: patch size\r\n\r\n* fix: runs now\r\n\r\n* fix: back to mlp\r\n\r\n* fix: stage 3?\r\n\r\n* fix: try again\r\n\r\n* fix: remove pos embed\r\n\r\n* fix: wtf\r\n\r\n* feat: mean pool test again?\r\n\r\n* feat: augments\r\n\r\n* fix: try no checkpointing\r\n\r\n* feat: 3 epoch augmentation train\r\n\r\n* fix: no randaugment\r\n\r\n* fix: dataset size\r\n\r\n* feat: 65k run with augs\r\n\r\n* fix: imagenet path\r\n\r\n* feat: try resume training multinode\r\n\r\n* fix: hostfile\r\n\r\n* fix: no flip for this train\r\n\r\n* fix: imagenet\r\n\r\n* refactor: remove unused\r\n\r\n* refactor: rename text_encoder -> nomic_encoder\r\n\r\n* refactor: remove captioner\r\n\r\n* chore: bump pydantic >= 2.0.0\r\n\r\n* feat: eval for clip models\r\n\r\n* feat: v1.5 config\r\n\r\n* fix: hf code\r\n\r\n* refactor: move hf tests to separate\r\n\r\n* chore: remove unused\r\n\r\n* refactor: remove\r\n\r\n* refactor: unused code\r\n\r\n* refactor: not used\r\n\r\n* fix: remove unused\r\n\r\n* refactor: remove xattn\r\n\r\n* refactor: remove xattn\r\n\r\n* fix: try to resume\r\n\r\n* fix: v1.5\r\n\r\n* fix: remove unused import\r\n\r\n* fix: remove ema\r\n\r\n* fix: remove ema\r\n\r\n* fix: instructions\r\n\r\n* feat: tracing code\r\n\r\n* feat: add stacks\r\n\r\n* feat: export_stacks=True\r\n\r\n* fix: with_stack\r\n\r\n* fix: tensorboard profiling (kind of) working\r\n\r\n* fix: don't profile, test full thing\r\n\r\n* feat: moar batch\r\n\r\n* feat: train\r\n\r\n* refactor: clean up code\r\n\r\n* feat: download data\r\n\r\n* fix: pydantic, workers crashing\r\n\r\n* fix: prefix\r\n\r\n* chore: ignore data folder\r\n\r\n* feat: loadable hf model\r\n\r\n* fix: map pooling bug\r\n\r\n* fix: comment old pooling\r\n\r\n* feat: flickr eval running\r\n\r\n* feat: flickr to config\r\n\r\n* feat: flickr eval train\r\n\r\n* fix: flickr eval doesn't hang\r\n\r\n* feat: biencoder test\r\n\r\n* fix: enforce no dynamic ntk\r\n\r\n* feat: unidirectional\r\n\r\n* feat: base timm models\r\n\r\n* fix: simplify vit pos_embed\r\n\r\n* fix: cls token confusion\r\n\r\n* feat: timm dinov2 with registers\r\n\r\n* wip vit rotary\r\n\r\n* feat: yolo 65k scratch vit\r\n\r\n* fix: hostfile\r\n\r\n* fix: revert back to bidirectional\r\n\r\n* fix: spelling\r\n\r\n* fix: path\r\n\r\n* fix: wandb\r\n\r\n* fix: shards\r\n\r\n* fix: reqs\r\n\r\n* feat: eva-style models, timm vit-base\r\n\r\n* fix: timm vit-b 224 image\r\n\r\n* feat: timm vit-b-16 first experiment\r\n\r\n* fix: no flip\r\n\r\n* feat: eva02 vit base\r\n\r\n* feat: pooling heads from timm vit\r\n\r\n* feat: add augreg vits as option\r\n\r\n* fix: remove pooling heads\r\n\r\n* fix: dumb renaming of model so eva loads with autoconfig\r\n\r\n* feat: eva config for training\r\n\r\n* fix: model loading\r\n\r\n* feat: 65k eva 3 epoch train\r\n\r\n* feat: map no clamp\r\n\r\n* fix: hostfile\r\n\r\n* fix: reduce workers\r\n\r\n* fix: no clamp\r\n\r\n* fix: config\r\n\r\n* feat: v1.5 train\r\n\r\n* fix: hostfile + config\r\n\r\n* fix: config for lower lr\r\n\r\n* fix: hamming\r\n\r\n* fix: train\r\n\r\n* feat: hf vision model code\r\n\r\n* fix: hostfile\r\n\r\n* fix: path\r\n\r\n* refactor: clean up code base\r\n\r\n* refactor: rename\r\n\r\n* fix: remove hostfile\r\n\r\n* refactor: remove sugarcrepe\r\n\r\n* style: black and isort\r\n\r\n* docs: readme and config fixes\r\n\r\n* fix: trainers, come back later","shortMessageHtmlLink":"feat: image-text (#3)"}},{"before":"4ed6743d3f724660db25e3fa9f6bfe5c8bcc61c3","after":"c545be2bc0074ea343eaa5247706965e13c1b5fa","ref":"refs/heads/main","pushedAt":"2024-05-21T19:07:32.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"feat: eval locov1 (#16)","shortMessageHtmlLink":"feat: eval locov1 (#16)"}},{"before":"20f395409bb759708c6c0310b9cd2ae91583db3d","after":"4ed6743d3f724660db25e3fa9f6bfe5c8bcc61c3","ref":"refs/heads/main","pushedAt":"2024-05-05T19:07:39.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"end to end training bug fixes (#15)\n\n* docs: add feature\r\n\r\n* fix: remove\r\n\r\n* docs: add feature\r\n\r\n* feat: matryoshka (#2)\r\n\r\n* feat: hamming + eval hamming\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka eval\r\n\r\n* fix: hf model load\r\n\r\n* feat: matryoshka loss weights\r\n\r\n* fix: don't set rotary to none since passed to config\r\n\r\n* fix: multi epoch\r\n\r\n* fix: errors if not present\r\n\r\n* fix: name\r\n\r\n* fix: rotary set\r\n\r\n* fix: rope scaling after 8192 in hf config\r\n\r\n* fix: return dict for sentence transformers\r\n\r\n* fix: config naming + eval scripts\r\n\r\n* fix: redundant if/else\r\n\r\n* style: black isort\r\n\r\n* feat: matryoshka for hf implementation\r\n\r\n* fix: eval with normalized scores for matryoshka\r\n\r\n* fix: matryoshka train with norm embeddings\r\n\r\n* refactor: rename\r\n\r\n* fix: config\r\n\r\n* fix: remove\r\n\r\n* fix: grad cache updates (#4)\r\n\r\n* fix: config uncomment\r\n\r\n* fix: rotary emb base fix\r\n\r\n* feat: hostfile\r\n\r\n* fix: config\r\n\r\n* fix: batch size\r\n\r\n* fix: lr schedule\r\n\r\n* fix: eval with correct step\r\n\r\n* fix: hostfile\r\n\r\n* fix: hf nomic code\r\n\r\n* fix: ordering of kwargs\r\n\r\n* fix: config\r\n\r\n* fix: config\r\n\r\n* fix: load state dict doesn't need to pass strict anymore\r\n\r\n* feat: convert to hf script\r\n\r\n* refactor: clean up clip loss to use openclip style loss\r\n\r\n* refactor: remove unused code\r\n\r\n* feat: first pass no grad cache train\r\n\r\n* fix: rank for multinode torchrun\r\n\r\n* fix: import\r\n\r\n* fix: set backend to nccl by default\r\n\r\n* fix: multinode download on all machines\r\n\r\n* fix: local rank instead of global rank\r\n\r\n* feat: multinode contrastive\r\n\r\n* fix: try without trainable logit scale\r\n\r\n* fix: try with gradient checkpointing\r\n\r\n* feat: deepspeed, lets try\r\n\r\n* fix: don't do cpu adam\r\n\r\n* fix: grad cache, lets see how long it'll take to train\r\n\r\n* fix: smaller chunk size\r\n\r\n* feat: grad cache exp, log to wandb\r\n\r\n* fix: chunk size\r\n\r\n* fix: chunk size\r\n\r\n* fix: unset for deepspeed testing\r\n\r\n* fix: smol data\r\n\r\n* feat: cpu offload\r\n\r\n* fix: add wandb and grad cache back\r\n\r\n* fix: data\r\n\r\n* fix: don't use token for hf code\r\n\r\n* fix: evals for mean pooling + push to hub\r\n\r\n* fix: reqs for evals\r\n\r\n* fix: remove pdb","shortMessageHtmlLink":"end to end training bug fixes (#15)"}},{"before":"8f5a5ce6cd92bc709f0426bbf8d2bbc73e24481f","after":null,"ref":"refs/heads/clm","pushedAt":"2024-04-19T14:39:20.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"}},{"before":"f3a04362fa4e15293a6da6b99a7791b2e6a013a1","after":"20f395409bb759708c6c0310b9cd2ae91583db3d","ref":"refs/heads/main","pushedAt":"2024-04-19T14:39:19.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Merge pull request #34 from nomic-ai/clm\n\nfix: causal language contrastive fixes (#14)","shortMessageHtmlLink":"Merge pull request #34 from nomic-ai/clm"}},{"before":null,"after":"8f5a5ce6cd92bc709f0426bbf8d2bbc73e24481f","ref":"refs/heads/clm","pushedAt":"2024-04-19T14:37:59.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"fix: causal language contrastive fixes (#14)\n\n* feat: grad checkpointing for biencoder\r\n\r\n* fix: remove file if empty\r\n\r\n* fix: mteb for new version\r\n\r\n* fix: grad checkpointing + tokenizer\r\n\r\n* feat: pythia pretrain config\r\n\r\n* fix: uncomment data","shortMessageHtmlLink":"fix: causal language contrastive fixes (#14)"}},{"before":"e22c5f4b895c40cb2164bff6a9be3440bb766fb5","after":null,"ref":"refs/heads/update-public","pushedAt":"2024-03-26T14:31:44.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"}},{"before":"3ea6a661f1a929ec82e2a42d020abc8f5a9f7dd6","after":null,"ref":"refs/heads/local_ds","pushedAt":"2024-03-26T14:31:43.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"}},{"before":"f318113c240f427b035e1f2cfd223459a4247260","after":"f3a04362fa4e15293a6da6b99a7791b2e6a013a1","ref":"refs/heads/main","pushedAt":"2024-03-26T01:05:28.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Merge pull request #29 from nomic-ai/local_ds\n\nfeat: allow for local dataloading (#13)","shortMessageHtmlLink":"Merge pull request #29 from nomic-ai/local_ds"}},{"before":null,"after":"3ea6a661f1a929ec82e2a42d020abc8f5a9f7dd6","ref":"refs/heads/local_ds","pushedAt":"2024-03-26T01:05:04.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"feat: allow for local dataloading (#13)\n\n* feat: allow for local dataloading\r\n\r\n* style: isort black","shortMessageHtmlLink":"feat: allow for local dataloading (#13)"}},{"before":"3e26d2d9b4892a25f4b364a3c4fbf2873645b637","after":"f318113c240f427b035e1f2cfd223459a4247260","ref":"refs/heads/main","pushedAt":"2024-03-22T19:15:14.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Merge pull request #27 from nomic-ai/update-public\n\nfix: remove pdb (#12)","shortMessageHtmlLink":"Merge pull request #27 from nomic-ai/update-public"}},{"before":"324aec4af13f74a3501f196fef5529e96b95b224","after":"3e26d2d9b4892a25f4b364a3c4fbf2873645b637","ref":"refs/heads/main","pushedAt":"2024-03-22T19:13:36.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"feat: deepspeed mlm train and load working (#7) (#24)","shortMessageHtmlLink":"feat: deepspeed mlm train and load working (#7) (#24)"}},{"before":"3e26d2d9b4892a25f4b364a3c4fbf2873645b637","after":"324aec4af13f74a3501f196fef5529e96b95b224","ref":"refs/heads/main","pushedAt":"2024-03-22T19:10:08.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"fix: remove pdb (#12) (#26)","shortMessageHtmlLink":"fix: remove pdb (#12) (#26)"}},{"before":null,"after":"e22c5f4b895c40cb2164bff6a9be3440bb766fb5","ref":"refs/heads/update-public","pushedAt":"2024-03-22T19:09:51.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"fix: remove pdb (#12)","shortMessageHtmlLink":"fix: remove pdb (#12)"}},{"before":"d2293d8f113f93e19b9663d10266b45ee6bb0374","after":"3e26d2d9b4892a25f4b364a3c4fbf2873645b637","ref":"refs/heads/main","pushedAt":"2024-03-22T18:17:00.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"feat: deepspeed mlm train and load working (#7) (#24)","shortMessageHtmlLink":"feat: deepspeed mlm train and load working (#7) (#24)"}},{"before":"a52d8cacaa5b98f81623671612d8c1ff046eb824","after":"d2293d8f113f93e19b9663d10266b45ee6bb0374","ref":"refs/heads/main","pushedAt":"2024-03-22T16:58:36.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"fix: decoder finetuning + grad cache bugs (#21)\n\n* feat: matryoshka (#2)\r\n\r\n* feat: hamming + eval hamming\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka eval\r\n\r\n* fix: hf model load\r\n\r\n* feat: matryoshka loss weights\r\n\r\n* fix: don't set rotary to none since passed to config\r\n\r\n* fix: multi epoch\r\n\r\n* fix: errors if not present\r\n\r\n* fix: name\r\n\r\n* fix: rotary set\r\n\r\n* fix: rope scaling after 8192 in hf config\r\n\r\n* fix: return dict for sentence transformers\r\n\r\n* fix: config naming + eval scripts\r\n\r\n* fix: redundant if/else\r\n\r\n* style: black isort\r\n\r\n* feat: matryoshka for hf implementation\r\n\r\n* fix: eval with normalized scores for matryoshka\r\n\r\n* fix: matryoshka train with norm embeddings\r\n\r\n* refactor: rename\r\n\r\n* fix: config\r\n\r\n* docs: add feature\r\n\r\n* fix: remove\r\n\r\n* docs: add feature\r\n\r\n* fix: remove\r\n\r\n* docs: add feature\r\n\r\n* fix: grad cache updates (#4)\r\n\r\n* Fix contrastive grad cache (#5)\r\n\r\n* fix: grad cache updates\r\n\r\n* fix: uncomment\r\n\r\n* fix: config uncomment\r\n\r\n* fix: decoder finetuning + grad cache bugs (#6)\r\n\r\n* fix: eos token for list of negatives col\r\n\r\n* fix: getattr for models that don't have this\r\n\r\n* fix: rand state bug for grad cache","shortMessageHtmlLink":"fix: decoder finetuning + grad cache bugs (#21)"}},{"before":"7f81b9d29e41ea4b22d9ecece386e2c0b74b744f","after":"a52d8cacaa5b98f81623671612d8c1ff046eb824","ref":"refs/heads/main","pushedAt":"2024-02-23T16:12:49.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"fix: make hf sequence classification loadable (#12)\n\n* feat: matryoshka (#2)\r\n\r\n* feat: hamming + eval hamming\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka eval\r\n\r\n* fix: hf model load\r\n\r\n* feat: matryoshka loss weights\r\n\r\n* fix: don't set rotary to none since passed to config\r\n\r\n* fix: multi epoch\r\n\r\n* fix: errors if not present\r\n\r\n* fix: name\r\n\r\n* fix: rotary set\r\n\r\n* fix: rope scaling after 8192 in hf config\r\n\r\n* fix: return dict for sentence transformers\r\n\r\n* fix: config naming + eval scripts\r\n\r\n* fix: redundant if/else\r\n\r\n* style: black isort\r\n\r\n* feat: matryoshka for hf implementation\r\n\r\n* fix: eval with normalized scores for matryoshka\r\n\r\n* fix: matryoshka train with norm embeddings\r\n\r\n* refactor: rename\r\n\r\n* fix: config\r\n\r\n* docs: add feature\r\n\r\n* fix: remove\r\n\r\n* docs: add feature\r\n\r\n* fix: remove\r\n\r\n* docs: add feature\r\n\r\n* fix: grad cache updates (#4)\r\n\r\n* Fix contrastive grad cache (#5)\r\n\r\n* fix: grad cache updates\r\n\r\n* fix: uncomment\r\n\r\n* fix: config uncomment\r\n\r\n* fix: cleanup + fix hf sequence classification code","shortMessageHtmlLink":"fix: make hf sequence classification loadable (#12)"}},{"before":"70f7baf0dff4e0019776e8c98257a3ad3fa4d687","after":"7f81b9d29e41ea4b22d9ecece386e2c0b74b744f","ref":"refs/heads/main","pushedAt":"2024-02-20T22:26:41.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"fix: grad cache (#10)\n\n* feat: matryoshka (#2)\r\n\r\n* feat: hamming + eval hamming\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka eval\r\n\r\n* fix: hf model load\r\n\r\n* feat: matryoshka loss weights\r\n\r\n* fix: don't set rotary to none since passed to config\r\n\r\n* fix: multi epoch\r\n\r\n* fix: errors if not present\r\n\r\n* fix: name\r\n\r\n* fix: rotary set\r\n\r\n* fix: rope scaling after 8192 in hf config\r\n\r\n* fix: return dict for sentence transformers\r\n\r\n* fix: config naming + eval scripts\r\n\r\n* fix: redundant if/else\r\n\r\n* style: black isort\r\n\r\n* feat: matryoshka for hf implementation\r\n\r\n* fix: eval with normalized scores for matryoshka\r\n\r\n* fix: matryoshka train with norm embeddings\r\n\r\n* refactor: rename\r\n\r\n* fix: config\r\n\r\n* docs: add feature\r\n\r\n* fix: remove\r\n\r\n* docs: add feature\r\n\r\n* fix: remove\r\n\r\n* docs: add feature\r\n\r\n* fix: grad cache updates (#4)\r\n\r\n* Fix contrastive grad cache (#5)\r\n\r\n* fix: grad cache updates\r\n\r\n* fix: uncomment\r\n\r\n* fix: config uncomment","shortMessageHtmlLink":"fix: grad cache (#10)"}},{"before":"31e465efb0b76ffeb14ea0f1bebf7f3bc0b4d252","after":null,"ref":"refs/heads/zanussbaum-patch-1","pushedAt":"2024-02-12T20:12:37.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"}},{"before":"1e27692238fd259025c56b56926893d21f691d84","after":"70f7baf0dff4e0019776e8c98257a3ad3fa4d687","ref":"refs/heads/main","pushedAt":"2024-02-12T18:20:54.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Update README.md (#7)","shortMessageHtmlLink":"Update README.md (#7)"}},{"before":null,"after":"31e465efb0b76ffeb14ea0f1bebf7f3bc0b4d252","ref":"refs/heads/zanussbaum-patch-1","pushedAt":"2024-02-12T18:20:36.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Update README.md","shortMessageHtmlLink":"Update README.md"}},{"before":"38119dc484a9401c1e78f158d7ec2586fa37f6c1","after":"1e27692238fd259025c56b56926893d21f691d84","ref":"refs/heads/main","pushedAt":"2024-02-12T18:19:21.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"feat: matryoshka (#6)\n\n* feat: matryoshka (#2)\r\n\r\n* feat: hamming + eval hamming\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka embeddings\r\n\r\n* feat: matroyshka eval\r\n\r\n* fix: hf model load\r\n\r\n* feat: matryoshka loss weights\r\n\r\n* fix: don't set rotary to none since passed to config\r\n\r\n* fix: multi epoch\r\n\r\n* fix: errors if not present\r\n\r\n* fix: name\r\n\r\n* fix: rotary set\r\n\r\n* fix: rope scaling after 8192 in hf config\r\n\r\n* fix: return dict for sentence transformers\r\n\r\n* fix: config naming + eval scripts\r\n\r\n* fix: redundant if/else\r\n\r\n* style: black isort\r\n\r\n* feat: matryoshka for hf implementation\r\n\r\n* fix: eval with normalized scores for matryoshka\r\n\r\n* fix: matryoshka train with norm embeddings\r\n\r\n* refactor: rename\r\n\r\n* fix: config\r\n\r\n* docs: add feature\r\n\r\n* fix: remove","shortMessageHtmlLink":"feat: matryoshka (#6)"}},{"before":"e40ae617034623b0c91f3c205fb41af61635b9ad","after":"38119dc484a9401c1e78f158d7ec2586fa37f6c1","ref":"refs/heads/main","pushedAt":"2024-02-05T20:00:22.000Z","pushType":"pr_merge","commitsCount":3,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Merge pull request #4 from zanussbaum/trainer\n\nfeat: trainer class","shortMessageHtmlLink":"Merge pull request #4 from zanussbaum/trainer"}},{"before":"b0e94ca1bdff67c19f5be50d198458e83bab63e9","after":"e40ae617034623b0c91f3c205fb41af61635b9ad","ref":"refs/heads/main","pushedAt":"2024-02-05T15:31:39.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"zanussbaum","name":"Zach Nussbaum","path":"/zanussbaum","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/33707069?s=80&v=4"},"commit":{"message":"Merge pull request #2 from jxmorris12/patch-1\n\nfix: typo","shortMessageHtmlLink":"Merge pull request #2 from jxmorris12/patch-1"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAEXTNJigA","startCursor":null,"endCursor":null}},"title":"Activity ยท nomic-ai/contrastors"}