Gustaf Ahdritz
48256248b5
Remove non-serializable type from model config
2025-06-06 11:14:55 -07:00
Gustaf Ahdritz
4c5e51e4de
Add on-device initialization
2025-06-06 08:13:15 -07:00
Pedro Rodriguez
96d51b59d2
Open source weights! ( #97 )
...
Lint with Black / lint (push) Failing after 3s
Lint with isort / lint (push) Failing after 2s
Summary:
Add code to download weights and demo code for running model.
Weights at:
- https://huggingface.co/collections/facebook/blt-6801263d4ac1704702a192a6
- https://huggingface.co/facebook/blt
- https://huggingface.co/facebook/blt-1b
- https://huggingface.co/facebook/blt-7b
Test Plan:
2025-04-17 09:38:56 -07:00
Srinivasan Iyer
138c2f3494
Init distributed when loading model ( #94 )
...
Co-authored-by: Srini Iyer <sviyer@meta.com>
2025-04-08 13:57:28 -07:00
Pedro Rodriguez
b79eb3ef11
Get generation working for BLT ( #86 )
...
Lint with isort / lint (push) Failing after 3s
Lint with Black / lint (push) Failing after 3s
Summary:
Create a script for simple generation from BLT
Test Plan:
```
python -m bytelatent.generate_blt config=../internal-blt/configs/eval_blt.yaml
```
2025-04-01 16:07:55 -07:00
Pedro Rodriguez
7517ac2a9f
Get evals working again. ( #46 )
...
- PPL/validation: Works now and uses multi-gpu. For some reason 1 GPU differs from multi-GPU, can debug in a followup PR
- Generation evals likely work, but are very slow, so disabled for now
Test Plan:
```
torchrun --nproc-per-node 8 -m bytelatent.eval config=../internal-blt/configs/eval.yaml
```
2025-03-11 09:57:19 -07:00
Pedro Rodriguez
7044771a12
This includes fixes that make checkpointing and reloading work correctly. ( #35 )
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
It also batches in a first set of changes for fixing eval code
Summary:
Test Plan:
2025-01-27 16:56:42 -08:00
