NOTICE: float8_experimental has moved

We have moved float8_experimental to pytorch/ao

• import float8_experimental is now import torchao.float8

float8_experimental

This is an early version of a library for accelerating training with float8 in native PyTorch according to the recipes laid out in https://arxiv.org/pdf/2209.05433.pdf. The codebase strives to stay small, easily hackable, debuggable with native PyTorch tooling, and composable with key systems such as autograd, torch.compile and distributed. With torch.compile on, initial results show throughput speedups of up to 1.2x on small scale (8 GPUs) LLaMa pretraining jobs.

⚠️ See the feature tracker for upcoming features.

⚠️ Backwards compatibility is not guaranteed at this point. The codebase is in active development and will change rapidly.

installation

⚠️ For now, use the latest PyTorch nightly for best results with torch.compile.

pip install .

# Optionally install editable
pip install -e .

# Optionally Install dev tooling
pip install -e ".[dev]"

Single GPU User API

We provide two per-tensor scaling strategies: dynamic and delayed. See https://arxiv.org/pdf/2209.05433.pdf, Section 4.3 for more details. These strategies are configurable separately for activations (input), weights (weight) and gradients (grad_output).

float8 linear with dynamic scaling for input, weight and grad_output

This is the most accurate recipe as every tensor is scaled dynamically.

from float8_experimental import (
    convert_to_float8_training,
    precompute_float8_dynamic_scale_for_fsdp,
)

# create model
m = Model(...)

# optional: filter modules from being eligible for float8 conversion
def module_filter_fn(mod: torch.nn.Module, fqn: str):
    # don't convert the output module
    if fqn == "output":
        return False
    # don't convert linear modules with weight dimensions not divisible by 16
    if isinstance(mod, torch.nn.Linear):
        if mod.in_features % 16 != 0 or mod.out_features % 16 != 0:
            return False
    return True

# convert all `torch.nn.Linear` modules to `Float8Linear`
convert_to_float8_training(m, module_filter_fn=module_filter_fn)

# optional: use FSDP
model = FSDP(model, use_orig_params=True)

# optional: enable torch.compile for improved performance
m = torch.compile(m)

# toy training loop
for _ in range(N_ITER):
    optimizer.zero_grad()
    y = m(x)
    y.sum().backward()
    optimizer.step()

    # specific to fsdp2 + dynamic scaling, when fp8 all-gather is turned on
    # this method is optional but is highly recommended for performance
    # it calcuclates scales for all parameters in a single all-reduce
    precompute_float8_dynamic_scale_for_fsdp(model)

float8 linear with delayed scaling

This is theoretically the most performant recipe as it minimizes memory reads.

from float8_experimental import (
    convert_to_float8_training,
    sync_float8_amax_and_scale_history,
    ScalingType,
)

# create model
m = Model(...)

# optional: configure for compatibility with FSDP. Note that workarounds
# gated with config.enable_amax_init and
# config.enable_pre_and_post_forward are needed for
# autocast + compile + FSDP + float8 to work
from float8_experimental import Float8LinearConfig, ScalingType, CastConfig
config = Float8LinearConfig(
    enable_amax_init = False,  # only needed for autocast + compile + FSDP +  float8 delayed
    enable_pre_and_post_forward, False  # only needed for autocast + compile + FSDP +  float8 delayed
    cast_config_input=CastConfig(scaling_type=ScalingType.DELAYED),
    cast_config_weight=CastConfig(scaling_type=ScalingType.DELAYED),
    cast_config_grad_output=CastConfig(scaling_type=ScalingType.DELAYED),
)

# convert all `torch.nn.Linear` modules to `Float8Linear`, specifying scaling
# type
convert_to_float8_training(
    m,
    config=config,
)

# optional: use FSDP
model = FSDP(model, use_orig_params=True)

# optional: enable torch.compile for improved performance
m = torch.compile(m)

# toy training loop
for _ in range(N_ITER):
    optimizer.zero_grad()
    y = m(x)
    y.sum().backward()

    # specific to float8 with delayed scaling: separate step to sync scales/amaxes
    # in the future, this may move to a context manager
    sync_float8_amax_and_scale_history(model)

    optimizer.step()

Multi GPU User API

We compose with the DTensor based distributed APIs, such as FSDP, TP and SP. Please see the torchtitan repository for e2e examples on using float8_experimental in a distributed setting.

Testing

# run single-GPU unit tests
pytest test/test_base.py

# run single-GPU compile tests
pytest test/test_compile.py

# run single-GPU numerics integration tests
pytest test/test_numerics_integration.py

# run a two-GPU integration test on FSDP
./test/test_fsdp.sh

# run integration tests on the DTensor TP/SP integration
./test/test_dtensor.sh

# run integration tests on the FSDP2 integration
python test/test_fsdp2/test_fsdp2.py

# run all of these tests
./test/test_everything.sh

Benchmarking

# benchmark the torch._scaled_mm function on LLaMa 2 70B shapes
./benchmarks/bench_matmul.py

# benchmark fw/bw of `Linear` and `Float8Linear` on LLaMa 2 70B shapes
# make sure to turn on torch.compile to get the best performance
./benchmarks/bench_linear_float8.py -o ../tmp/test.txt --compile

License

PyTorch has a BSD 3-Clause License, as found in the LICENSE file.