-
Notifications
You must be signed in to change notification settings - Fork 6
Issues: nntile/nntile
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.
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
Author
Label
Projects
Milestones
Assignee
Sort
Issues list
Bug: BeamSearch of LLM inference does not take into account temperature parameter when choosing best beams
bug
Something isn't working
#201
opened Nov 13, 2024 by
Muxas
Bug: LlamaAttention bias gradients seem to be computed incorrectly
bug
Something isn't working
#156
opened Aug 16, 2024 by
Muxas
Bug: check if amount Flops, counted at Something isn't working
src/starpu/flash_*.cc
are correct
bug
#154
opened Aug 13, 2024 by
Muxas
Test: double check that
test_backward
functions actually check gradient over input
#129
opened Jul 31, 2024 by
Muxas
Test: training loss curve of a minimalistic custom Llama model against PyTorch
ci
Builds, testing, images, and so on
#126
opened Jul 29, 2024 by
Muxas
Test: rework Embedding layer testing
enhancement
New feature or request
good first issue
Good for newcomers
#112
opened Jul 21, 2024 by
Muxas
Add NNTile dtypes into Python namespace
api
Public programming interfaces
enhancement
New feature or request
#102
opened Jul 14, 2024 by
Muxas
Bug: total_sum_accum on CUDA is broken with bf16 type
bug
Something isn't working
#95
opened Jul 11, 2024 by
Muxas
Logger: show number of read-write operations per second just like FLOPs
api
Public programming interfaces
enhancement
New feature or request
logger
Tensorboard-compatible logger
#74
opened Jul 7, 2024 by
Muxas
Code design: make C++ non-template class Tensor with dynamic data type field
api
Public programming interfaces
#54
opened Jun 24, 2024 by
Muxas
Models: study U-net and U-net++ models
model-zoo
Support a new model or make a showcase
#49
opened Jun 16, 2024 by
Muxas
Logger: draw distribution of tasks along computational resources
enhancement
New feature or request
logger
Tensorboard-compatible logger
#48
opened Jun 16, 2024 by
Muxas
Support popular encoder-decoder models like T5
model-zoo
Support a new model or make a showcase
#44
opened May 29, 2024 by
daskol
Bug: GPT2 model does not train if some tiles are partial
bug
Something isn't working
#41
opened May 29, 2024 by
Muxas
Operation: a single-level Strassen matrix multiplication
api
Public programming interfaces
enhancement
New feature or request
#39
opened May 28, 2024 by
Muxas
Operation: split "add_fiber" into an in-place "add_fiber_i" and out-of-place "add_fiber" operations
api
Public programming interfaces
good first issue
Good for newcomers
#37
opened May 28, 2024 by
Muxas
6 tasks
Operation: split "add" into an in-place "add_i" and out-of-place "add" operations
api
Public programming interfaces
good first issue
Good for newcomers
#35
opened May 28, 2024 by
Muxas
6 tasks
Docs: move all Python single-string explanation of functions into docstring
documentation
Improvements or additions to documentation
#32
opened May 28, 2024 by
Muxas
Model: add GPT-Neo and GPT-Neo-X
enhancement
New feature or request
model-zoo
Support a new model or make a showcase
#27
opened May 28, 2024 by
Muxas
Previous Next
ProTip!
Type g p on any issue or pull request to go back to the pull request listing page.