Layers as clusters, money as arcs, brand marks on the biggest stars
AI Intelligence Graph
Starting the view
Follows the graph · 1 entity
Full file coming soon. PyTorch's links, deals and figures below are what the graph holds today.Not collected yet: 360, Money, Investors, Market share, Legal, News and events, Earnings calls. Anthropic is the first complete company file (people, investors, legal, hiring, birth to now); other companies follow one at a time. Open Anthropic's file
Overview
Ask the graph
answers only from the graph's sourced facts, with citationsForecast record
No forecast published here yet. Each one names its basis and the date it resolves, is scored afterwards, and the hit rate shows here in public once the first resolves.
L0 Labs · Standard / protocol · USA
PyTorch
The default training/inference framework
- Links
- 12
- Deals from filings
- 0
- Exposed if it fails
- 3
- 35 within 2 steps
- It depends on
- 3
- 3 within 2 steps
No company file yet: revenue, customers and other figures appear here once they are taken from its filings.
What it is
PyTorch (governed by the PyTorch Foundation under the Linux Foundation, originated at Meta) is the substrate nearly every frontier model is built on, sitting above CUDA and below the labs. It is a genuine standards-layer dependency that binds labs to Nvidia via the CUDA backend, and its device-abstraction is the battleground for AMD/TPU portability.
Pick a link to move the graph there: · · · · · · ·
Connections12 links
Pulled 28 Oct 2026 · AI Systems Map links
Its suppliers
3Who depends on it
3Rivals
3Designs
1Compare two: what PyTorch shares with another entity
Pick a second entity to see their direct links, the suppliers, backers, rivals and regulators they share, and the deals with the same parties. The graph lights the same set.
People and hiring
Pulled 28 Oct 2026, 01:29 EDT · GitHub releases · data to 24 Oct 2026
Posts and releases
official channels only · newest first- 24 Oct 2026PyTorchtensordict v0.15.3: TensorDict v0.15.3 ↗GitHub releaseHighlights - print(td) shows the compact form by default, and notebooks show a foldable tree of dtypes and shapes. set printoptions(compact=False) restores the detailed form ( 3143, 3106). - New ways to save and exchange data: safetensors f
- 19 Oct 2026PyTorchtensordict v0.15.2: TensorDict v0.15.2 ↗GitHub releaseHighlights TensorDict 0.15.2 supports torch 2.11 and 2.12 again, fixes 54 bugs, adds 8 features, and speeds up the calls that TorchRL, rsl rl and verl make every step. 24 calls now warn ahead of changes in TensorDict 0.17, and tensordict no
- 17 Oct 2026PyTorchtensordict v0.15.1: TensorDict v0.15.1 ↗GitHub releaseHighlights TensorDict 0.15.1 is a patch release with 46 bug fixes. They cover update , copy and in-place math on lazy stacks, h5 files and TypedTensorDict , UnbatchedTensor and non-tensor data, NumPy conversion, tensordict.nn , and torch.co
- 15 Oct 2026PyTorchtensordict v0.15.0: TensorDict v0.15.0 ↗GitHub releaseTensorDict v0.15.0 TensorDict 0.15.0 contains the 581 pull requests merged since 0.14.0, 37 of which also shipped in the 0.14.1 to 0.14.3 patch releases: - tensordict is pure Python: one 690 KB wheel for every platform and Python version. -
- 8 Oct 2026PyTorchtensordict v0.14.3: TensorDict v0.14.3 ↗GitHub releaseHighlights TensorDict 0.14.3 is a patch release with 24 bug fixes. They cover indexing with boolean masks and mixed advanced indices, memory-mapped loading and archives, TensorDictStore , UnbatchedTensor and non-tensor data, tensorclasses,
- 30 Sep 2026PyTorchpytorch v2.14.1: PyTorch 2.14.1 Release ↗GitHub releaseThis release is meant to fix the following regressions and silent correctness issues: Silent correctness fixes - Fix incorrect torch.linalg.lstsq solutions on MPS for complex batched underdetermined systems ( 196113), fixed by 196128 - Fix
- 3 Sep 2026PyTorchtorchtitan v0.3.0 ↗GitHub releaseWhat's changed Compared with v0.2.2, TorchTitan 0.3.0 adds typed Python configuration, expands model and declarative parallelism support, and enhances MoE execution, graph-native training, training efficiency, SFT, and RL. Configuration arc
- 2 Sep 2026PyTorchpytorch v2.14.0: PyTorch 2.14.0 Release ↗GitHub releasePyTorch 2.14.0 Release Notes - Highlights - Backwards Incompatible Changes - Deprecations - New Features - Improvements - Bug fixes - Performance - Documentation - Security - Developers Highlights <table <tr <td <strong NVGEMM</strong bring
- 22 Jul 2026PyTorchFBGEMM v1.8.0: FBGEMM_GPU v1.8.0 ↗GitHub releaseFBGEMM GPU v1.8.0 Release Notes Highlights New DRAM+SSD Composite Backend Introduced DRAM SSD backend type enabling DRAM as L1 cache with SSD backing store for embeddings ( 5774, 5802, 5926) Full training pipeline integration with metadata
- 8 Jul 2026PyTorchpytorch v2.13.0: PyTorch 2.13.0 Release ↗GitHub releasePyTorch 2.13.0 Release Notes - Highlights - Backwards Incompatible Changes - Deprecations - New Features - Improvements - Bug fixes - Performance - Documentation - Developers Highlights <table <tr <td <strong FlexAttention</strong lands on
- 18 Jun 2026PyTorchpytorch v2.12.1: PyTorch 2.12.1 Release, bug fix release ↗GitHub releaseThis release is meant to fix the following regressions and silent correctness issues: Regression fixes - Fix nondeterministic outputs in test batch invariance with FLASH ATTN on NVIDIA B200 GPUs ( 181248), fixed by updating Triton to 3.7.1
- 18 May 2026PyTorchFBGEMM v1.7.0: FBGEMM_GPU v1.7.0 Release Notes ↗GitHub releaseHighlights Inference & Production Deployment TurboSSDInferenceModule with streaming updates and snapshot loading for HSTU serving ( 5558, 5554) AMD/ROCm support for SSD TBE inference with cache locking and dedicated memcpy streams ( 5559, 5
No people in public roles recorded here yet.
More platforms: not collected yet
Release notes come from each company's GitHub organisation. People enter only in public roles, from filings, company pages, talks and papers. Blogs, talks, papers and executives' public posts come next, from official channels; LinkedIn and X only through licensed datasets. No scraping of personal profiles.
Data sources2 sources
| Source | Feeds | Data dated | Pulled | Pulled again | Status |
|---|---|---|---|---|---|
| GitHub releases20 release notes | People and hiring | 24 Oct 2026 | 28 Oct 2026, 01:29 EDT | daily | fresh |
| AI Systems Map links12 links; built July 2026, deals re-checked against filings 27 to 28 Oct 2026 | Connections | · | 28 Oct 2026 | when the map is reviewed | hand-built; deals checked against filings |
Times are New York time. Free sources are pulled on the schedule shown by the pipeline runner (data/corpus/run_pipelines.py), which logs every run; paid sources run only with the owner's yes; filings and newsroom pages are re-read when a new document is filed.
Links come from the AI Systems Map ontology. Deals carry the filing's own wording and a quote, checked against the document; latest source 2026-10-26. Not investment advice.