099d30f814
add a super small crappy glob for windows that only matches a single unique file. this will make CI happy but we can't train on sharded data on windows until this is improved
2024-05-22 13:08:18 +00:00
05be4f6d82
readme changes
2024-05-22 12:32:25 +00:00
edb0df967a
continued changes for sharded dataloader
2024-05-22 11:59:09 +00:00
31310282e1
extend dataloader to be sharded
2024-05-21 21:33:24 +00:00
967420d1d1
fix print format warning for size_t vs int
2024-05-21 19:30:02 +00:00
587506d096
torch tensor can't handle uint16 so let's convert to int32, which is silly because we'll convert to .long right after but ok
2024-05-21 19:22:09 +00:00
170c92b771
Merge pull request #440 from karpathy/feature/organize_data
...
refactor datasets
you'll have to re-generate the dataset .bin files, for example, `python dev/data/tinyshakespeare.py`.
see main README section "major changes log" for more details.
2024-05-21 11:32:38 -07:00
54ccbd300c
docs on master-breaking changes around dataset file representation
2024-05-21 18:31:19 +00:00
7d58fd2abb
adjust py file as well and make the errors better
2024-05-21 18:29:55 +00:00
f7cb77f3d9
docs on master-breaking change around how we store data .bin files
2024-05-21 18:20:51 +00:00
ccc240caab
make comment more helpful
2024-05-21 18:01:35 +00:00
d53608820a
the write_datafile function accepts python list, which i think is faster but i didn't check
2024-05-21 17:55:56 +00:00
9bad49a99a
also fix the python file. that should be it now, w.r.t. the new token format .bin files
2024-05-21 17:46:14 +00:00
666145e7b8
also fix the .c file and fp32 file
2024-05-21 17:30:56 +00:00
1defbd4a19
adjust the dataloader to load the new .bin data files, and both prod datasets to use it instead
2024-05-21 17:05:27 +00:00
f7cbb30316
step 2 of dataloader refactor: separate out the dataloader to its own file, change its signature a little bit, and (notably) change from cudaMallocHost to a simple malloc, so that this file doesn't have to be cuda aware
2024-05-21 15:32:46 +00:00
4bbd01fe12
add fineweb, and add the first version of a new write_shard function that contains a header properly
2024-05-21 00:16:07 +00:00
c51cd70cfb
fix a slip in gitignore, i think i am getting tired today
2024-05-20 23:22:31 +00:00
bced34d71e
adjust gitignore
2024-05-20 23:15:44 +00:00
7d11b7996c
i also forgot to include the readme file for the new dev/data dir
2024-05-20 23:13:42 +00:00
71774b3c3d
oops forgot to include data_common.py
2024-05-20 23:10:49 +00:00
f671cf9288
more changes, trying to help people out because when this merges to master it will brick everyone's code...
2024-05-20 23:02:41 +00:00
722e5b2fe5
refactor how we treat datasets, because we're about to have more of them and we don't want them to clutter up root dir etc. this is only step 1, i'm about to refactor a bunch of the dataloading, how the .bin files work and are loaded, how the DataLoader works, etc. This is all needed to support good evals and training at scale
2024-05-20 22:43:40 +00:00
6c8bc172cc
Merge pull request #436 from ChrisDryden/boundscheck
...
Moved bounds checks to outside of the kernel
2024-05-20 00:41:41 +01:00
2751fa0880
Merge pull request #435 from ChrisDryden/warpsize
...
Added warpsize as a constant for better compile time optimization and standardization
2024-05-20 00:39:15 +01:00
6de1137e0e
Moved bounds checks outside of kernel into assertions
2024-05-19 23:17:45 +00:00
ead5d35973
Added warpsize as a constant for better compile time optimization and standardization
2024-05-19 22:54:47 +00:00
1809740a1c
Merge pull request #433 from junaire/jun/some_fix
...
Check if file exists using platform specific commands & Add *.o to gi…
2024-05-19 21:02:04 +01:00
00c291c679
Merge pull request #434 from ngc92/profile-fix
...
initialize multi_gpu_config so profile doesn't crash
2024-05-19 20:57:11 +01:00
299ce659ec
initialize multi_gpu_config so profile doesn't crash
2024-05-19 22:25:00 +03:00
5a26799efb
Merge branch 'ngc92-gradient-clipping'
2024-05-19 17:51:44 +00:00
bc58cd1dc1
fix small bugs in grad clip, introduce a GPU CPU synch point to communicate the float grad_clip to print it, and small printing changes
2024-05-19 17:51:29 +00:00
c2d12f725e
small touchups to grad clip
2024-05-19 17:07:55 +00:00
da9f4e21b0
Merge branch 'gradient-clipping' of https://github.com/ngc92/llm.c into ngc92-gradient-clipping
2024-05-19 16:03:53 +00:00
9e645314e6
Check if file exists using platform specific commands & Add *.o to gitignore
...
Signed-off-by: Jun Zhang <jun@junz.org >
2024-05-20 00:00:21 +08:00
4a899f5685
Merge pull request #431 from karpathy/feature/hellaswag
...
add hellaswag and mmlu PyTorch reference eval
2024-05-19 16:53:38 +01:00
8d55c4a6ac
fix a bug, we have to be careful to make sure we evaluate loss at the token just before the first completion token, because that is the prediction for the first token, and its accuracy
2024-05-19 15:45:28 +00:00
4e14b52285
report both acc and acc_norm
2024-05-19 15:18:04 +00:00
7bd2389c1f
add mmlu as well and refine both a bit
2024-05-19 14:23:51 +00:00
aec9ce5d0a
move hellaswag file to dev
2024-05-19 12:18:02 +00:00
77b991281f
add hellaswag reference eval. it looks pretty bad honestly, the examples themselves, but i'm told it is predictive... hmm
2024-05-19 11:51:01 +00:00
66ce5766e0
fixed up dev/cuda
2024-05-18 23:06:44 +03:00
589ead1e4f
updated code to adapt to latest changes
2024-05-18 22:47:42 +03:00
c3a3b9daa5
fixed profile target
2024-05-18 22:47:40 +03:00
a9947a8315
added a flag and the missing sqrt;
...
testing now has clipping enabled
2024-05-18 22:47:15 +03:00
d7a81ef26f
added a useful mixed precision utility for dev/cuda
2024-05-18 22:45:44 +03:00
44d45bdd6a
first draft for gradient clipping by global norm
2024-05-18 22:45:21 +03:00
3de05baaf4
Merge branch 'ngc92-errchk'
2024-05-18 18:35:28 +00:00
4374360015
adjust wte upper bound a bit, and print always because this part is really tricky and i don't trust anything other than manual inspection, even if we pass, allegedly
2024-05-18 18:34:24 +00:00
bf36a4b6b5
improved numerical error checking:
...
tighter tolarances
relative tolerance based of bf16 epsilon
less verbose output if all is OK
2024-05-18 13:27:35 +03:00
Andrej Karpathy