Skip to content

Issues: ahrefs/ocannl

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

80 Open 219 Closed
80 Open 219 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

XLA backend based directly on the assignments interface
#300 opened Dec 23, 2024 by lukstafi non-essential-backend
Write a paper for OCaml or ML Workshop (ideally 2025); explain why OCANNL has dimensions units but not axis labels enhancement Priority below unlabeled
#299 opened Dec 20, 2024 by lukstafi
Rename "dimensions label" -> "dimensions units", make dimensions units more usable enhancement Priority below unlabeled
#298 opened Dec 20, 2024 by lukstafi
Implement a BERT model, replicate ModernBERT to use ModernBERT weights explore Priority below "enhancement", non-blocking for milestones
#297 opened Dec 20, 2024 by lukstafi
Write / flesh-out lowering_and_inlining.md and audit low_level.ml
#296 opened Dec 18, 2024 by lukstafi v0.7
Consider allowing Shared_cross_streams to be written to from multiple streams.
#294 opened Dec 10, 2024 by lukstafi
@lukstafi
Implement sharding and slicing with minimal copying
#293 opened Dec 10, 2024 by lukstafi
@lukstafi
Consider measuring the overhead of creating (recording) events enhancement Priority below unlabeled
#292 opened Nov 15, 2024 by lukstafi
Audit manually specifying sharing in a tensor node memory mode
#291 opened Nov 14, 2024 by lukstafi
The use of shared_merge_buffer should depend on the tensor node's memory_mode (on all backends) bug Highest priority
#290 opened Nov 10, 2024 by lukstafi
@lukstafi
Make memory reporting for CUDA more meaningful
#289 opened Oct 16, 2024 by lukstafi v0.7
Consider introducing proper static verification of merge buffer nodes, but "in the right direction" enhancement Priority below unlabeled
#288 opened Oct 11, 2024 by lukstafi v0.8
Support half precision in the GCCJIT backend
#284 opened Sep 11, 2024 by lukstafi non-essential-backend
When in a label the same identifier repeats consecutively (e.g. ident 3 times), let debug_name collapse it (e.g. ident3)
#281 opened Aug 30, 2024 by lukstafi v0.9
ppx_op: Include config.label in the label of the primary (returned) tensor, when available, and not incorporating a label from a function argument enhancement Priority below unlabeled
#280 opened Aug 30, 2024 by lukstafi v0.9
1
Example training loop using DisTrO low-communication distributed data parallelism explore Priority below "enhancement", non-blocking for milestones
#278 opened Aug 27, 2024 by lukstafi
Anything we can learn from krnl and autograph? explore Priority below "enhancement", non-blocking for milestones
#277 opened Jul 30, 2024 by lukstafi
Replicate Andrej Karpathy's "LLM101n: Let's build a Storyteller" explore Priority below "enhancement", non-blocking for milestones
#275 opened Jul 24, 2024 by lukstafi
Support quantization for optimizers: low-bit optimizers explore Priority below "enhancement", non-blocking for milestones
#271 opened Jul 6, 2024 by lukstafi
Any lessons from Imbue for training-in-the-large? explore Priority below "enhancement", non-blocking for milestones
#270 opened Jul 4, 2024 by lukstafi
1
Take a look at Tiramisu Polyhedral Compiler explore Priority below "enhancement", non-blocking for milestones
#267 opened May 27, 2024 by lukstafi
Consider implementing a Cranelift backend (CPU) explore Priority below "enhancement", non-blocking for milestones
#266 opened May 25, 2024 by lukstafi
Study Candle -- a minimalistic Rust framework explore Priority below "enhancement", non-blocking for milestones
#265 opened May 25, 2024 by lukstafi
Consider implementing Lean Attention (Flash Attention + softmax-as-reduce) explore Priority below "enhancement", non-blocking for milestones
#263 opened May 21, 2024 by lukstafi
@lukstafi
Superoptimizers for tensor programs explore Priority below "enhancement", non-blocking for milestones
#261 opened May 11, 2024 by lukstafi
ProTip! Updated in the last three days: updated:>2024-12-22.