Pedro Rodriguez
f912535cb7
Merge 655eca670d
into sapling-pr-archive-EntilZha
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
2025-02-14 15:46:06 -08:00
Pedro Rodriguez
88dedaa2ec
Merge a3e0647d03
into sapling-pr-archive-EntilZha
2025-02-14 15:45:43 -08:00
Pedro Rodriguez
655eca670d
Minimal working eval
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
Summary:
Test Plan:
2025-02-14 23:45:29 +00:00
Pedro Rodriguez
a3e0647d03
Make apex logs less noisy
...
Summary:
Test Plan:
2025-02-14 23:45:28 +00:00
Pedro Rodriguez
52590842e0
merge commit for archive created by Sapling
2025-02-14 22:51:24 +00:00
Pedro Rodriguez
f94babc94e
Make it possible to specify multiple config files
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
Summary:
Make it possible to specify multiple config files.
Parsing CLI is not a special case anymore, just uses the same config inheritance method.
Test Plan:
Test that this iterpolates in the right order via unit tests
Sample usage, loads the internal config, which references bytelatent/configs/entropy_model.yaml. The precendence order is:
- Default pydantic args
- Included configs, eg `config`
- CLI args
```
python -m bytelatent.print_config config=internal/configs/entropy_model.yaml eval=null
```
Summary:
Test Plan:
2025-02-14 22:50:57 +00:00
Pedro Rodriguez
018bf98798
Merge aa78c96ea4
into sapling-pr-archive-EntilZha
2025-02-14 13:06:55 -08:00
Pedro Rodriguez
aa78c96ea4
Make it possible to specify multiple config files
...
Summary:
Make it possible to specify multiple config files.
Parsing CLI is not a special case anymore, just uses the same config inheritance method.
Test Plan:
Test that this iterpolates in the right order via unit tests
Sample usage, loads the internal config, which references bytelatent/configs/entropy_model.yaml. The precendence order is:
- Default pydantic args
- Included configs, eg `config`
- CLI args
```
python -m bytelatent.print_config config=internal/configs/entropy_model.yaml eval=null
```
2025-02-14 21:06:50 +00:00
Pedro Rodriguez
ed6300375f
Merge bec0164820
into sapling-pr-archive-EntilZha
2025-02-14 13:04:04 -08:00
Pedro Rodriguez
bec0164820
Make it possible to specify multiple config files
...
Summary:
Test Plan:
Test that this iterpolates in the right order, config -> configs -> cli args
```
# All three sources
python -m bytelatent.print_config config=bytelatent/configs/debug.yaml configs=[internal/configs/s3_debug.yaml] eval=null
# What worked before
python -m bytelatent.print_config config=internal/configs/s3_debug.yaml eval=null
```
2025-02-14 21:03:57 +00:00
Pedro Rodriguez
1c7031b4c4
Merge be3ff12cfe
into sapling-pr-archive-EntilZha
2025-02-14 13:03:38 -08:00
Pedro Rodriguez
be3ff12cfe
Make it possible to specify multiple config files
...
Summary:
Test Plan:
Test that this iterpolates in the right order, config -> configs -> cli args
```
# All three sources
python -m bytelatent.print_config config=bytelatent/configs/debug.yaml configs=[internal/configs/s3_debug.yaml] eval=null
# What worked before
python -m bytelatent.print_config config=internal/configs/s3_debug.yaml eval=null
```
2025-02-14 21:03:26 +00:00
Srinivasan Iyer
f3e8125f74
using apex rmsnorm ( #57 )
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
* using apex rmsnorm
* added message for missing apex
* black
* missed a print
---------
Co-authored-by: Srini Iyer <sviyer@meta.com>
2025-02-14 11:22:03 -08:00
Srinivasan Iyer
c49e25171e
Update README.md ( #58 )
2025-02-14 11:16:49 -08:00
Pedro Rodriguez
8c61ab5e67
Fix multiprocessing dataloader checkpointing and use it in the train script ( #50 )
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
2025-02-13 11:58:23 -08:00
Pedro Rodriguez
84afa0f121
merge commit for archive created by Sapling
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
2025-02-13 19:01:55 +00:00
Pedro Rodriguez
53529dcc78
Fix multiprocessing dataloader checkpointing and use it in the train script
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
Summary:
Test Plan:
2025-02-13 19:01:49 +00:00
Pedro Rodriguez
76e7b001bb
Merge 0c6cb995a0
into sapling-pr-archive-EntilZha
2025-02-13 10:39:03 -08:00
Pedro Rodriguez
0c6cb995a0
Fix multiprocessing dataloader checkpointing and use it in the train script
...
Summary:
Test Plan:
2025-02-13 18:38:58 +00:00
Pedro Rodriguez
85c2f28f26
Test first batch matches ( #53 )
...
Summary:
Test Plan:
2025-02-13 10:05:08 -08:00
Pedro Rodriguez
45d52b7ae3
Merge ab8f8a4412
into sapling-pr-archive-EntilZha
2025-02-13 10:04:43 -08:00
Pedro Rodriguez
ab8f8a4412
Test first batch matches
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
Summary:
Test Plan:
2025-02-13 18:04:30 +00:00
Srinivasan Iyer
9d907fed1c
disable reshard after forward ( #56 )
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
Co-authored-by: Srini Iyer <sviyer@meta.com>
2025-02-12 18:33:53 -08:00
Srinivasan Iyer
48e4ad0bd2
make sure max_encoder_seq_length matches ( #55 )
...
* make sure max_encoder_seq_length matches
* black and assert comment
---------
Co-authored-by: Srini Iyer <sviyer@meta.com>
2025-02-12 18:27:22 -08:00
Pedro Rodriguez
078791996f
Merge ece82cb960
into sapling-pr-archive-EntilZha
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
2025-02-12 11:25:18 -08:00
Pedro Rodriguez
ece82cb960
Make it possible to specify multiple config files
...
Summary:
Test Plan:
Test that this iterpolates in the right order, config -> configs -> cli args
```
# All three sources
python -m bytelatent.print_config config=bytelatent/configs/debug.yaml configs=[internal/configs/s3_debug.yaml] eval=null
# What worked before
python -m bytelatent.print_config config=internal/configs/s3_debug.yaml eval=null
```
2025-02-12 19:25:06 +00:00
Pedro Rodriguez
15d9c40abe
Merge 3e3193c1d4
into sapling-pr-archive-EntilZha
2025-02-12 10:24:54 -08:00
Pedro Rodriguez
c0c5bdba91
Merge c54c9f0517
into sapling-pr-archive-EntilZha
2025-02-12 10:24:45 -08:00
Pedro Rodriguez
3e3193c1d4
Fix multiprocessing dataloader checkpointing and use it in the train script
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
Summary:
Test Plan:
2025-02-12 18:24:40 +00:00
Pedro Rodriguez
c54c9f0517
Test first batch matches
...
Summary:
Test Plan:
2025-02-12 18:24:39 +00:00
Pedro Rodriguez
ec59c13d81
Merge bd3cf61bb9
into sapling-pr-archive-EntilZha
2025-02-12 10:11:44 -08:00
Pedro Rodriguez
9613e0ea5f
merge commit for archive created by Sapling
2025-02-12 18:09:31 +00:00
Pedro Rodriguez
bd3cf61bb9
Fix multiprocessing dataloader checkpointing and use it in the train script
...
Summary:
Test Plan:
2025-02-12 18:09:26 +00:00
Pedro Rodriguez
4cee32ea8c
Test first batch matches
...
Summary:
Test Plan:
2025-02-12 18:09:26 +00:00
Pedro Rodriguez
b61a612bbb
Merge 92af9b3f56
into sapling-pr-archive-EntilZha
2025-02-12 10:07:50 -08:00
Pedro Rodriguez
92af9b3f56
Test first batch matches
...
Summary:
Test Plan:
2025-02-12 18:07:22 +00:00
Pedro Rodriguez
c6cbacc8c1
merge commit for archive created by Sapling
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
2025-02-11 22:56:32 +00:00
Pedro Rodriguez
38cc67a953
Fix multiprocessing dataloader checkpointing and use it in the train script
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
Summary:
Test Plan:
2025-02-11 22:56:26 +00:00
Pedro Rodriguez
c4b7a01b2b
Merge 5c8fb4f1b3
into sapling-pr-archive-EntilZha
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
2025-02-07 15:27:12 -08:00
Pedro Rodriguez
5c8fb4f1b3
Fix multiprocessing dataloader checkpointing and use it in the train script
...
Summary:
Test Plan:
2025-02-07 23:27:05 +00:00
Srinivasan Iyer
22c7fe1d1c
fix save and reload model state ( #49 )
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
Co-authored-by: Srini Iyer <sviyer@meta.com>
2025-02-07 14:27:47 -08:00
Pedro Rodriguez
fe45f69fbf
Add bpb and n_bytes to metric logging ( #41 )
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
Summary:
Test Plan:
2025-02-07 13:14:30 -08:00
Pedro Rodriguez
b35206d756
merge commit for archive created by Sapling
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
2025-02-07 21:13:42 +00:00
Pedro Rodriguez
8d7338308e
Add bpb and n_bytes to metric logging
...
Lint with Black / lint (push) Has been cancelled
Lint with isort / lint (push) Has been cancelled
Summary:
Test Plan:
2025-02-07 21:13:36 +00:00
Pedro Rodriguez
f783846574
merge commit for archive created by Sapling
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
2025-02-07 00:26:06 +00:00
Pedro Rodriguez
b6396eb0f4
Add bpb and n_bytes to metric logging
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
Summary:
Test Plan:
2025-02-07 00:26:01 +00:00
Srinivasan Iyer
aebdc481a8
Fix init and repro ( #48 )
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
* Fix init and repro
* comment + black
---------
Co-authored-by: Srini Iyer <sviyer@meta.com>
2025-02-06 14:18:02 -08:00
Pedro Rodriguez
ab594996a9
Merge 4e2ed0aa05
into sapling-pr-archive-EntilZha
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
2025-02-06 10:24:36 -08:00
Pedro Rodriguez
4e2ed0aa05
Add bpb and n_bytes to metric logging
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
Summary:
Test Plan:
2025-02-06 18:08:01 +00:00
Pedro Rodriguez
936d9437be
Allow ArrowIterator to read from json ( #45 )
...
Lint with Black / lint (push) Waiting to run
Lint with isort / lint (push) Waiting to run
Summary:
Currently, arrow iterator can only read arrow files. However, the pyarrow library can read
other formats, including jsonlines. This allows the same ArrowIterator to read from jsonlines,
so we can read from the original source data, and simply omit the entropy column when doing so
Test Plan:
Run train script until dataloader starts
2025-02-06 09:57:22 -08:00