torch<2.15.0,>=2.13.0
packaging==26.0
huggingface_hub>=1.31.0
peft==0.21.2
tokenizers==0.23.3
transformers==5.20.1
accelerate==1.15.0
datasets==5.0.1
trl==1.14.1
hf_xet>=1.5.2
kernels<0.18.0,>=0.17.2
trackio==0.40.0
typing-extensions==4.15.0
optimum==1.16.2
hf_transfer
sentencepiece
gradio==6.29.1
pydantic==2.13.5
addict
fire
PyYAML==6.0.3
requests
wandb
einops
colorama
numba<1.0,>=0.63.1
numpy<3.0,>=2.1
typer==0.25.1
evaluate==0.4.1
scipy
nvidia-ml-py==12.560.30
art
tensorboard
python-dotenv==1.0.1
s3fs==2025.10.0
gcsfs==2025.10.0
adlfs==2026.2.0
ocifs==1.3.2
zstandard==0.22.0
fastcore
lm_eval==0.4.11
langdetect==1.0.9
immutabledict==4.2.0
antlr4-python3-runtime==4.13.2
schedulefree==1.4.1
openenv-core>=0.1.0
axolotl-contribs-lgpl==0.0.7
axolotl-contribs-mit==0.0.6
posthog==6.7.11
mistral-common==1.12.0

[:platform_machine != "aarch64"]
fla-core==0.5.2
flash-linear-attention==0.5.2

[:sys_platform != "darwin"]
bitsandbytes==0.50.2
triton>=3.4.0
liger-kernel==0.8.4

[:sys_platform != "darwin" and platform_machine != "aarch64"]
xformers>=0.0.33.post2
torchao==0.18.0

[apollo]
apollo-torch

[auto-gptq]
auto-gptq==0.5.1

[baseten]
truss==0.18.30

[coreweave]
boto3>=1.35

[deepspeed]
deepspeed<0.20.0,>=0.18.6
deepspeed-kernels

[fbgemm-gpu]
fbgemm-gpu-genai>=1.3.0

[fla]
flash-linear-attention[tilelang]==0.5.2
tilelang==0.1.14

[galore]
galore_torch

[kernels-curated]
kernels[curated]

[llmcompressor]
llmcompressor>=0.10.0

[mamba-ssm]
mamba-ssm==1.2.0.post1
causal_conv1d

[mlflow]
mlflow

[modal]
modal==1.3.0.post1

[opentelemetry]
opentelemetry-api
opentelemetry-sdk
opentelemetry-exporter-prometheus
prometheus-client

[optimizers]
galore_torch
apollo-torch
lomo-optim==0.1.1
torch-optimi==0.2.1
came_pytorch==0.1.3
scao==0.2.2
q-galore-torch==1.0

[ray]
ray[train]>=2.59.0

[ringmaster]
axolotl-ringmaster>=0.2.6

[vllm]
vllm>=0.30.0
