Skip to content

Commit bc5fbf3

Browse files
kmehantkmehant
committed
feat: update requirements:
Signed-off-by: Mehant Kammakomati <[email protected]>
1 parent bb3c88e commit bc5fbf3

File tree

5 files changed

+72
-118
lines changed

5 files changed

+72
-118
lines changed

‎plugins/accelerated-peft/requirements.txt‎

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,8 @@
55
accelerate >= 0.29
66

77
# bitsandbytes for the BNB plugin
8-
# exact version is needed 0.45.1 for torch upgrade to 2.6
8+
# versions above 0.45.1 to support torch 2.6
9+
# exact version is used since upper bound is not known
910

1011
bitsandbytes == 0.45.1
1112

‎sample-configurations/moe-scattermoe-granite-ep8-padding-free-foak-sample-configuration.yaml‎

Lines changed: 0 additions & 51 deletions
This file was deleted.

‎scripts/benchmarks/compare_with_reference.py‎

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -10,16 +10,16 @@
1010

1111
# default columns to compare
1212
DEFAULT_PLOT_COLUMNS = [
13-
# "mem_torch_mem_alloc_in_bytes",
14-
# "mem_peak_torch_mem_alloc_in_bytes",
13+
"mem_torch_mem_alloc_in_bytes",
14+
"mem_peak_torch_mem_alloc_in_bytes",
1515
'mem_nvidia_mem_reserved',
1616
"train_loss",
1717
"train_tokens_per_second",
1818
]
1919
# Used as combined identifier of experiment
2020
DEFAULT_INDICES = [
2121
"framework_config",
22-
# "peft_method",
22+
"peft_method",
2323
"model_name_or_path",
2424
"num_gpus",
2525
"per_device_train_batch_size",
@@ -30,7 +30,7 @@
3030
"train_runtime",
3131
"train_steps_per_second",
3232
"train_samples_per_second",
33-
# "mem_nvidia_mem_reserved",
33+
"mem_nvidia_mem_reserved",
3434
]
3535

3636
DEFAULT_REFERENCE_FILEPATH = "scripts/benchmarks/refs/a100_80gb.csv"

‎scripts/benchmarks/refs/requirements_moe.txt‎

Lines changed: 64 additions & 60 deletions
Original file line numberDiff line numberDiff line change
@@ -1,87 +1,91 @@
1-
accelerate==1.0.1
2-
aiohappyeyeballs==2.4.6
3-
aiohttp==3.11.12
1+
accelerate==1.6.0
2+
aiohappyeyeballs==2.6.1
3+
aiohttp==3.12.12
44
aiosignal==1.3.2
5-
attrs==25.1.0
6-
bitsandbytes==0.43.3
7-
certifi==2025.1.31
8-
charset-normalizer==3.4.1
9-
contourpy==1.3.1
5+
attrs==25.3.0
6+
bitsandbytes==0.45.1
7+
causal-conv1d==1.5.0.post8
8+
certifi==2025.4.26
9+
charset-normalizer==3.4.2
10+
contourpy==1.3.2
1011
cycler==0.12.1
11-
datasets==2.21.0
12+
datasets==3.6.0
1213
dill==0.3.8
1314
einops==0.8.1
14-
filelock==3.17.0
15+
filelock==3.18.0
1516
flash_attn==2.7.4.post1
16-
-e git+https://github.com/foundation-model-stack/fms-acceleration.git@21af5fb9f2989b3dbf443c016e4c0470b536a593#egg=fms_acceleration&subdirectory=plugins/framework
17-
-e git+https://github.com/foundation-model-stack/fms-acceleration.git@21af5fb9f2989b3dbf443c016e4c0470b536a593#egg=fms_acceleration_aadp&subdirectory=plugins/attention-and-distributed-packing
18-
-e git+https://github.com/foundation-model-stack/fms-acceleration.git@21af5fb9f2989b3dbf443c016e4c0470b536a593#egg=fms_acceleration_foak&subdirectory=plugins/fused-ops-and-kernels
19-
-e git+https://github.com/foundation-model-stack/fms-acceleration.git@21af5fb9f2989b3dbf443c016e4c0470b536a593#egg=fms_acceleration_moe&subdirectory=plugins/accelerated-moe
20-
-e git+https://github.com/foundation-model-stack/fms-acceleration.git@21af5fb9f2989b3dbf443c016e4c0470b536a593#egg=fms_acceleration_peft&subdirectory=plugins/accelerated-peft
21-
fms-hf-tuning @ git+https://github.com/foundation-model-stack/fms-hf-tuning.git@fdc7527510692ada03e4303df1549cebc5139b31
22-
fonttools==4.56.0
23-
frozenlist==1.5.0
24-
fsspec==2024.6.1
25-
huggingface-hub==0.29.0
17+
-e git+https://github.com/kmehant/fms-acceleration.git@bb3c88ecf960403295c34a49b9a4acdeff782d1d#egg=fms_acceleration&subdirectory=plugins/framework
18+
-e git+https://github.com/kmehant/fms-acceleration.git@bb3c88ecf960403295c34a49b9a4acdeff782d1d#egg=fms_acceleration_aadp&subdirectory=plugins/attention-and-distributed-packing
19+
-e git+https://github.com/kmehant/fms-acceleration.git@bb3c88ecf960403295c34a49b9a4acdeff782d1d#egg=fms_acceleration_foak&subdirectory=plugins/fused-ops-and-kernels
20+
-e git+https://github.com/kmehant/fms-acceleration.git@bb3c88ecf960403295c34a49b9a4acdeff782d1d#egg=fms_acceleration_moe&subdirectory=plugins/accelerated-moe
21+
-e git+https://github.com/kmehant/fms-acceleration.git@bb3c88ecf960403295c34a49b9a4acdeff782d1d#egg=fms_acceleration_peft&subdirectory=plugins/accelerated-peft
22+
fms-hf-tuning @ git+https://github.com/foundation-model-stack/fms-hf-tuning.git@9ca5739a4f7c1f0c9446b28d1bb80939d4199a75
23+
fonttools==4.58.2
24+
frozenlist==1.7.0
25+
fsspec==2025.3.0
26+
hf-xet==1.1.3
27+
huggingface-hub==0.33.0
2628
idna==3.10
27-
Jinja2==3.1.5
28-
kernel-hyperdrive @ git+https://github.com/fabianlim/kernel-hyperdrive.git@45036497e12444ca98a6f0072204538aee4543ba
29+
Jinja2==3.1.6
2930
kiwisolver==1.4.8
3031
llvmlite==0.44.0
32+
mamba-ssm==2.2.4
3133
markdown-it-py==3.0.0
3234
MarkupSafe==3.0.2
33-
matplotlib==3.10.0
35+
matplotlib==3.10.3
3436
mdurl==0.1.2
3537
mpmath==1.3.0
36-
multidict==6.1.0
38+
multidict==6.4.4
3739
multiprocess==0.70.16
38-
networkx==3.4.2
39-
numba==0.61.0
40+
networkx==3.5
41+
ninja==1.11.1.4
42+
numba==0.61.2
4043
numpy==1.26.4
41-
nvidia-cublas-cu12==12.1.3.1
42-
nvidia-cuda-cupti-cu12==12.1.105
43-
nvidia-cuda-nvrtc-cu12==12.1.105
44-
nvidia-cuda-runtime-cu12==12.1.105
44+
nvidia-cublas-cu12==12.4.5.8
45+
nvidia-cuda-cupti-cu12==12.4.127
46+
nvidia-cuda-nvrtc-cu12==12.4.127
47+
nvidia-cuda-runtime-cu12==12.4.127
4548
nvidia-cudnn-cu12==9.1.0.70
46-
nvidia-cufft-cu12==11.0.2.54
47-
nvidia-curand-cu12==10.3.2.106
48-
nvidia-cusolver-cu12==11.4.5.107
49-
nvidia-cusparse-cu12==12.1.0.106
49+
nvidia-cufft-cu12==11.2.1.3
50+
nvidia-cufile-cu12==1.11.1.6
51+
nvidia-curand-cu12==10.3.5.147
52+
nvidia-cusolver-cu12==11.6.1.9
53+
nvidia-cusparse-cu12==12.3.1.170
5054
nvidia-cusparselt-cu12==0.6.2
51-
nvidia-nccl-cu12==2.20.5
55+
nvidia-nccl-cu12==2.21.5
5256
nvidia-nvjitlink-cu12==12.4.127
53-
nvidia-nvtx-cu12==12.1.105
54-
packaging==24.2
55-
pandas==2.2.3
56-
peft==0.13.2
57-
pillow==11.1.0
58-
propcache==0.2.1
59-
protobuf==5.29.3
57+
nvidia-nvtx-cu12==12.4.127
58+
packaging==25.0
59+
pandas==2.3.0
60+
peft==0.14.0
61+
pillow==11.2.1
62+
propcache==0.3.2
63+
protobuf==5.29.5
6064
psutil==7.0.0
61-
pyarrow==19.0.1
65+
pyarrow==20.0.0
6266
Pygments==2.19.1
63-
pyparsing==3.2.1
67+
pyparsing==3.2.3
6468
python-dateutil==2.9.0.post0
65-
pytz==2025.1
69+
pytz==2025.2
6670
PyYAML==6.0.2
6771
regex==2024.11.6
68-
requests==2.32.3
69-
rich==13.9.4
70-
safetensors==0.5.2
72+
requests==2.32.4
73+
rich==14.0.0
74+
safetensors==0.5.3
7175
sentencepiece==0.2.0
72-
setuptools==75.8.0
73-
simpleeval==0.9.13
76+
setuptools==80.9.0
77+
simpleeval==1.0.3
7478
six==1.17.0
7579
sympy==1.13.1
76-
threadpoolctl==3.5.0
77-
tokenizers==0.21.0
78-
torch==2.4.1
80+
threadpoolctl==3.6.0
81+
tokenizers==0.21.1
82+
torch==2.6.0
7983
tqdm==4.67.1
80-
transformers==4.49.0
81-
triton==3.0.0
82-
trl==0.14.0
83-
typing_extensions==4.12.2
84-
tzdata==2025.1
85-
urllib3==2.3.0
84+
transformers==4.52.4
85+
triton==3.2.0
86+
trl==0.17.0
87+
typing_extensions==4.14.0
88+
tzdata==2025.2
89+
urllib3==2.4.0
8690
xxhash==3.5.0
87-
yarl==1.18.3
91+
yarl==1.20.1

‎tox.ini‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -34,7 +34,7 @@ commands =
3434
# some models need this for tokenizers
3535
pip install protobuf
3636

37-
# for mamba based models
37+
# install mamba kernels to benchmark mamba based models
3838
pip install --no-build-isolation mamba_ssm[causal-conv1d]>=2.0.0
3939

4040
# install the plugins for test
@@ -45,7 +45,7 @@ commands =
4545
python -m fms_acceleration.cli install -e {toxinidir}/plugins/accelerated-moe
4646

4747
# install the flash attn at the last
48-
pip install flash-attn --no-build-isolation
48+
pip install flash-attn
4949

5050
# run the benchmark script
5151
bash scripts/run_benchmarks.sh {posargs:"1 2" "4 8" benchmark_outputs}

0 commit comments

Comments
 (0)