Add files using upload-large-folder tool

README.md

@@ -0,0 +1,8 @@
+# open-source
+
+Local copies of [open-thoughts/OpenThoughts3-1.2M](https://huggingface.co/datasets/open-thoughts/OpenThoughts3-1.2M), kept under the `open-source` folder.
+
+- `open-source/hub/`: original parquet shards and the upstream dataset card. License: Apache-2.0.
+- `open-source/datasets/`: local Hugging Face `datasets` arrow cache of the same data.
+
+This is a backup of files already on this machine. The upstream dataset remains open-thoughts/OpenThoughts3-1.2M.

open-source/datasets/open_thoughts3-1.2_m-train-00034-of-00112.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2df58a96f21f31d763281afcf3b47bd2e8ea74fecfc427b92056b713a84f6f02
+size 543756168

open-source/datasets/open_thoughts3-1.2_m-train-00040-of-00112.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa4a1adeb917ad548f43f24f5bb1439c40bc9a60fdf5b96c998da8a933e986d3
+size 541488832

open-source/datasets/open_thoughts3-1.2_m-train-00051-of-00112.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f18a988eb0879cc7874f4b83415c1abfbf4a6bdc8f9db8e6a005de1584bf9d68
+size 540247984

open-source/hub/README.md

@@ -0,0 +1,124 @@
+---
+dataset_info:
+  features:
+  - name: difficulty
+    dtype: int64
+  - name: source
+    dtype: string
+  - name: domain
+    dtype: string
+  - name: conversations
+    list:
+    - name: from
+      dtype: string
+    - name: value
+      dtype: string
+  splits:
+  - name: train
+    num_bytes: 59763369750
+    num_examples: 1200000
+  download_size: 28188197544
+  dataset_size: 59763369750
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train-*
+license: apache-2.0
+task_categories:
+- text-generation
+tags:
+- reasoning
+- mathematics
+- code
+- science
+library_name: datasets
+---
+
+<p align="center">
+    <img src="https://huggingface.co/datasets/open-thoughts/open-thoughts-114k/resolve/main/open_thoughts.png" width="50%">
+</p>
+
+<p align="center">
+<a href="https://arxiv.org/abs/2506.04178" style="margin-right: 24px;">paper</a> |
+<a href="https://huggingface.co/datasets/open-thoughts/OpenThoughts3-1.2M" style="margin-right: 24px; margin-left: 24px;">dataset</a> |
+<a href="https://huggingface.co/open-thoughts/OpenThinker3-7B" style="margin-left: 24px;">model</a>
+</p>
+
+> [!NOTE]
+> We have released a paper for OpenThoughts! See our paper [here](https://arxiv.org/abs/2506.04178).
+
+<a href="https://github.com/bespokelabsai/curator/">
+ <img src="https://huggingface.co/datasets/bespokelabs/Bespoke-Stratos-17k/resolve/main/made_with_curator.png" alt="Made with Curator" width=200px>
+</a>
+
+
+# OpenThoughts3-1.2M
+
+## Dataset Description
+
+- **Homepage:** https://www.open-thoughts.ai/
+- **Repository:** https://github.com/open-thoughts/open-thoughts
+
+Open-source state-of-the-art reasoning dataset with 1.2M rows. 🚀
+
+[OpenThoughts3-1.2M](https://arxiv.org/abs/2506.04178) is the third iteration in our line of [OpenThoughts](https://openthoughts.ai) datasets, building on our previous [OpenThoughts-114k](https://huggingface.co/datasets/open-thoughts/OpenThoughts-114k) and [OpenThoughts2-1M](https://huggingface.co/datasets/open-thoughts/OpenThoughts2-1M).
+This time around, we scale even further and generate our dataset in a much more systematic way -- [OpenThoughts3-1.2M](https://arxiv.org/abs/2506.04178) is the result of a rigorous experimental pipeline, that ablates over design choices surrounding question sourcing and selection, as well as answer generation. 
+The final dataset consists of 850,000 math questions, 250,000 code questions, and 100,000 science questions. Annotations are done with QwQ-32B.
+
+This dataset was used to train [OpenThinker3-7B](https://huggingface.co/open-thoughts/OpenThinker3-7B), which beats **all** similarly sized open-data reasoning models.
+
+See our [paper](https://arxiv.org/abs/2506.04178) and [blog post](https://www.open-thoughts.ai/blog/ot3) for more details. OpenThinker3-32B to follow! 👀
+
+
+# OpenThinker3-7B Model Performance
+
+Our [OpenThinker3-7B](https://huggingface.co/open-thoughts/OpenThinker3-7B) model trained on this dataset is the state-of-the-art open-data reasoning dataset at its scale. 
+
+We conduct all our evaluations using [Evalchemy](https://github.com/mlfoundations/evalchemy). In the table below, we bold values in each column that are within 2 standard errors of the best. 
+
+| Model                                                                                           | Data  | AIME24 | AIME25 |  AMC23 | MATH500 | HMMT O2/25 | LCB 06/24-01/25 | CodeElo | CodeForces | GPQA-D | JEEBench |
+| ----------------------------------------------------------------------------------------------- | ----- | ------ | ------ | ------ | ------- | ---------- | --------------- | ------- | ---------- | ------ | -------- |
+| [OpenThinker-7B](https://huggingface.co/open-thoughts/OpenThinker-7B)                           | ✅    |  30.7  |  22.0  |  72.5  |   82.8  |   15.7     |    26.1         |  11.1   |  14.9      |  38.6  |  45.3    |
+| [OpenThinker2-7B](https://huggingface.co/open-thoughts/OpenThinker2-7B)                         | ✅    |  60.7  |  38.7  |  89.8  |   87.6  |   24.7     |    40.6         |  22.8   |  26.6      |  47.0  |  65.1    |
+| **[OpenThinker3-7B](https://huggingface.co/open-thoughts/OpenThinker3-7B)**                     | ✅    |**69.0**|**53.3**|**93.5**| **90.0**|   **42.7** |    **51.7**     |  31.0   |**32.2**    |  53.7  |**72.4**  |
+| [DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B) | ❌    |  51.3  |  38.0  |  92.0  |   88.0  |   25.0     |    34.5         |  19.9   |  21.1      |  33.2  |  50.4    |
+| [OpenR1-Distill-7B](https://huggingface.co/open-r1/OpenR1-Distill-7B)                           | ✅    |  57.7  |  39.7  |  87.0  |   88.0  |   25.7     |    30.7         |  30.1   |  29.3      |**58.9**|  68.7    |
+| [Llama-3.1-Nemotron-Nano-8B-v1](https://huggingface.co/nvidia/Llama-3.1-Nemotron-Nano-8B-v1)    | ✅    |  62.0  |  48.0  |**94.0**|   89.4  |   26.7     |    **50.9**     |  30.9   |**32.9**    |  52.9  |  70.7    |
+| [AceReason-Nemotron-7B](https://huggingface.co/nvidia/AceReason-Nemotron-7B)                    | ✅    |**71.0**|  50.7  |**93.8**|   89.8  |   33.3     |    44.3         |**32.9** |**30.9**    |  52.9  |  64.3    |
+
+# OpenThoughts3 Data Curation and Scaling Recipe
+
+Please see our [paper](https://arxiv.org/abs/2506.04178) for full details and experimental results.
+
+[OpenThoughts3-1.2M](https://huggingface.co/datasets/open-thoughts/OpenThoughts3-1.2M) was created through the following steps:
+
+0. Run our 1000+ ablations and find the best datasets, filtering strategies, and annotation strategies.  
+1. 🙋 **Source questions** from the best strategies from Step 0. These question strategies can be both human-written (e.g., StackExchange Physics) or synthetic (e.g., [ai2-adapt-dev/openmath-2-math](https://huggingface.co/datasets/ai2-adapt-dev/openmath-2-math)).
+2. 🧹 **Filter** the questions down to 180k math, 60k code, 60k science.
+3. 🧹 **Deduplicate** datasets. 
+4. 🧹 **Downsample** so that we have 75k questions in total.
+5. 💡 **Annotate each question 16x** with QwQ-32B, arriving at a 1.2M dataset.
+
+![ot3-datacuration-pipeline](ot3-datacuration-pipeline.png)
+
+
+# Links
+- 📝 [OpenThoughts Paper](https://arxiv.org/abs/2506.04178)
+- 📊 [OpenThoughts3-1.2M and OpenThinker3-7B Blog Post](https://www.open-thoughts.ai/blog/ot3)
+- 💻 [Open Thoughts GitHub Repository](https://github.com/open-thoughts/open-thoughts)
+- 🧠 [OpenThoughts3-1.2M dataset](https://huggingface.co/datasets/open-thoughts/OpenThoughts3-1.2M) - this dataset.
+- 🤖 [OpenThinker3-7B model](https://huggingface.co/open-thoughts/OpenThinker3-7B)
+
+# Citation
+```
+@misc{guha2025openthoughtsdatarecipesreasoning,
+  title={OpenThoughts: Data Recipes for Reasoning Models}, 
+  author={Etash Guha and Ryan Marten and Sedrick Keh and Negin Raoof and Georgios Smyrnis and Hritik Bansal and Marianna Nezhurina and Jean Mercat and Trung Vu and Zayne Sprague and Ashima Suvarna and Benjamin Feuer and Liangyu Chen and Zaid Khan and Eric Frankel and Sachin Grover and Caroline Choi and Niklas Muennighoff and Shiye Su and Wanjia Zhao and John Yang and Shreyas Pimpalgaonkar and Kartik Sharma and Charlie Cheng-Jie Ji and Yichuan Deng and Sarah Pratt and Vivek Ramanujan and Jon Saad-Falcon and Jeffrey Li and Achal Dave and Alon Albalak and Kushal Arora and Blake Wulfe and Chinmay Hegde and Greg Durrett and Sewoong Oh and Mohit Bansal and Saadia Gabriel and Aditya Grover and Kai-Wei Chang and Vaishaal Shankar and Aaron Gokaslan and Mike A. Merrill and Tatsunori Hashimoto and Yejin Choi and Jenia Jitsev and Reinhard Heckel and Maheswaran Sathiamoorthy and Alexandros G. Dimakis and Ludwig Schmidt},
+  year={2025},
+  eprint={2506.04178},
+  archivePrefix={arXiv},
+  primaryClass={cs.LG},
+  url={https://arxiv.org/abs/2506.04178}, 
+}
+```

open-source/hub/data/train-00000-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f79e754b68b2c031bc8104ccc8a5fa4775f9d381ccb8fc87577ca7d5ef409434
+size 286338841

open-source/hub/data/train-00001-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e59a001bb6a7565240218c3b5344f35a0a1766d8ae9b2d9f7cf9b63f6b5f4875
+size 286016790

open-source/hub/data/train-00002-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fbd316949c5f4b2f26d4b19d4a684cbd002f434e4fee65bd735f7a7831a2879
+size 286884425

open-source/hub/data/train-00003-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e31191ea49ee8836c02844436ab4557d04953674adce65e6632cc5670d8a4dda
+size 286819062

open-source/hub/data/train-00008-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a78980d53a362754713a2e128e1bef47224c49b9938879f3a379e5e58a13bfd5
+size 286265825

open-source/hub/data/train-00009-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58ad5ab1c514d49574557a4e727ef8072b12bb2c480d2a43e44e7f5707018af7
+size 286697739

open-source/hub/data/train-00010-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2383c1cd08a4a91e1c610aa40e81a072cef2ffe6e2e36095117732202240bfca
+size 286378606

open-source/hub/data/train-00012-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a15574b4fdf1a43cc4866a8f9f1e5e6f5dd5ad3b5cbc346d14199d8ab74fdb7c
+size 285818419

open-source/hub/data/train-00014-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e2b89495a0ce5f358251d1c0d00c6e2ebd3fac8efa52dc101d15a77122ee355
+size 285730881

open-source/hub/data/train-00028-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0297636cc94e7a8d0f85e3a9d9149d6d571ba849d63d0ea20c6108f413dbf4b
+size 230240603

open-source/hub/data/train-00031-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4784ebd78eec60831f1197751ca147219274433fa0069b35fc489591aa45691a
+size 229986921

open-source/hub/data/train-00034-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15327a1e22617ccc79369e725b12b5fabe01795a4aa7c6600a04929badf19a6f
+size 229757683

open-source/hub/data/train-00041-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68ef1bfec408e7df1c21586c9968c29eccec3b30699549ada464c14786374b6a
+size 229840546

open-source/hub/data/train-00043-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91cf122c89dde017e11de49cd785620c0d620af5efb946f252c72fc79d6ed4a6
+size 230253672

open-source/hub/data/train-00044-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e6c4f1135ee75fd7a42d7f5fe05279c0a2a65da3fb02c46a37dba97b6b14ab7
+size 229521195

open-source/hub/data/train-00046-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee467dc37cd7deb4dd877a2d03f1eff6b7cbec5dfa688517ea55c1e975919798
+size 229514676

open-source/hub/data/train-00047-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4492b12672e1af34554aaf33e670246d3b02595b2af2bede63bfb4a5b9553d1
+size 229873686

open-source/hub/data/train-00052-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f82b9239054c027504e4178b2e259c713ef4200e53db457159f965bf9503bca
+size 229634289

open-source/hub/data/train-00053-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2724156826cb6b3c2006dea302925b57e2709cefbe4a5910755dda007a407c07
+size 230319250

open-source/hub/data/train-00057-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de6d4e7b916586019672261ac720b0f0bea3a792fbcbfe23e4adc554c65182fc
+size 229415188

open-source/hub/data/train-00060-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fb7cc88110a8229b26f69ecca59ad6dee7f17fbfcf0d851ac0381846b42b6cc
+size 230472872

open-source/hub/data/train-00064-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adbf547ebf0fe21ed99b227449f3e476101c746e16c38cecc9365ec8f22ab6a3
+size 229561500

open-source/hub/data/train-00068-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1e9c4e190613be524d298906c714a482068649cc3989a6c5995f656438d9baa
+size 229671059

open-source/hub/data/train-00070-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ea2dbfb49d2e39ddead9d51fc741a182e2cd9a2006c993c000cdc79cc01e688
+size 230476744

open-source/hub/data/train-00074-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cabcc8a2ed2a963395a7dbc3debc1a12b24b2364b5a8efef67c325d7aadee9e4
+size 229238533

open-source/hub/data/train-00075-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa2344d39694cb76c6f811b8af967225f9bacba854f91c754cbc936240237018
+size 229495659

open-source/hub/data/train-00076-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c84be4720af1612c83c2be9a8e3697baafd7be6b423dcec811f371a7b83ae1fb
+size 229540901

open-source/hub/data/train-00077-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f791460b49e51e5c7c307bf49b1f270507d0bfcdad6621b58dcea176396077a7
+size 229724042

open-source/hub/data/train-00083-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f75af2eafe10437068ee64b1b52cfc11c1b1891712b45842d8c9a9b770b2c617
+size 229747416

open-source/hub/data/train-00084-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e6cb8c011fb47d5ac082ded587287e9b1a9d06b6c39603ba91193549ba3a76c
+size 229392055

open-source/hub/data/train-00087-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd0a4b935a5bbec95e3ef40ee74c6680c0b2dd9acd1f760267549e0c785701d9
+size 229835133

open-source/hub/data/train-00090-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee3593a1e86b76c11749dd6c2bc33a56720d4881f13ec32af29d3cd280001591
+size 229981834

open-source/hub/data/train-00093-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64981bb76f266101506cbf150305673f19eee06087286b6dd7f7d9c1deebaaf8
+size 229392739

open-source/hub/data/train-00098-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a897280636d3945decbad7ea7dcfe5755ad5e9817b00cf9dea1efc10d974884
+size 229532291

open-source/hub/data/train-00102-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dff887fb8805c4100cf1d894d5b462f3923edd874d72bbf5b7ab1f69a45bb9cf
+size 229184717

open-source/hub/data/train-00103-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9736642ca1d172d04c1f59c3e22af0676b54a3da95a8e85e8c68caaf591baeb
+size 230004467

open-source/hub/data/train-00104-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c8b98c729836c4c3031332bb96d7354b555f193c119fe0cea8f5c33c6634731
+size 229789354

open-source/hub/data/train-00105-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27a7fb2c1bf5990a99c8a51a8db21fd4335860833ceac3d999a5c35dc5b0aaa8
+size 229550406

open-source/hub/data/train-00106-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:194e670526ed399096a8574b27fa9aaa170e87ee467b5d13e15a88f7a118b110
+size 229512647

open-source/hub/data/train-00109-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcf86b3727728b0c4ff4584a3f52af85f3c7eaeefb38cae22753390cf06f416c
+size 229735306

open-source/hub/data/train-00110-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e952a1f11c106a40746a954b76aa5f15fdbbb449da98e86841a8c4397f308c6
+size 149421537

open-source/hub/data/train-00113-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfeffb6833f56aa5df10c3ae693774c2e8a1a3ac3afe1efa6fc9e23847bb86ef
+size 149835667

open-source/hub/data/train-00114-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5718848283a825db3d6e884c31650fc49c0ff62ae02af96f42074f360ac9bb4d
+size 149432526

open-source/hub/data/train-00115-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a12b2d5c33b0b4cc980a8d29fe1dd60de5493967a0fa0159529dda422b3556a1
+size 149562230

open-source/hub/data/train-00119-of-00120.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59158062a8a5243744120093dc0764b988792b2e38148d6a04c6d6c2a02929b1
+size 149792515