Skip to content

Commit

Permalink
Generate: end-to-end compilation (#30788)
Browse files Browse the repository at this point in the history 
* mvp

* added test (a few models need fixes)

* fix a few test cases

* test nits

* harder test 😈

* revert changes in stablelm

* test with improved condition

* add todo

* tmp commit

* merged with main

* nits

* add todo

* final corrections

* add docs for generation compilation

* docs nits

* add  tip

* PR suggestions

* add more details to the compilation docs

* fix cache positions

* cache is now init in generate; update docs

* tag test as flaky

* docs

* post rebase make fixup and other nits

* remove unintended changes

* whisper (encoder-decoder) not supported

* move token default updates to ; add tests for token defaults

* push changes

* manual rebase

* chameleon doesn't support this

* fix test_static_cache_mha_mqa_gqa (broken in another PR)

* docs: dynamic is better with end-to-end compilation
  • Loading branch information
@gante
gante authored Jul 29, 2024
1 parent 4992889 commit 7ffe25f
Show file tree
Hide file tree
Showing 11 changed files with 282 additions and 100 deletions.
139 changes: 106 additions & 33 deletions docs/source/en/llm_optims.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -18,59 +18,109 @@ Basic inference is slow because LLMs have to be called repeatedly to generate th
This guide will show you how to use the optimization techniques available in Transformers to accelerate LLM inference.

> [!TIP]
> Hugging Face also provides [Text Generation Inference (TGI)](https://hf.co/docs/text-generation-inference), a library dedicated to deploying and serving highly optimized LLMs for inference. It includes more optimization features not included in Transformers, such as continuous batching for increasing throughput and tensor parallelism for multi-GPU inference.
> Hugging Face also provides [Text Generation Inference (TGI)](https://hf.co/docs/text-generation-inference), a library dedicated to deploying and serving highly optimized LLMs for inference. It includes deployment-oriented optimization features not included in Transformers, such as continuous batching for increasing throughput and tensor parallelism for multi-GPU inference.
## Static kv-cache and torch.compile
## Static kv-cache and `torch.compile`

During decoding, a LLM computes the key-value (kv) values for each input token and since it is autoregressive, it computes the same kv values each time because the generated output becomes part of the input now. This is not very efficient because you're recomputing the same kv values each time.

To optimize this, you can use a kv-cache to store the past keys and values instead of recomputing them each time. However, since the kv-cache grows with each generation step and is dynamic, it prevents you from taking advantage of [torch.compile](./perf_torch_compile), a powerful optimization tool that fuses PyTorch code into fast and optimized kernels.
To optimize this, you can use a kv-cache to store the past keys and values instead of recomputing them each time. However, since the kv-cache grows with each generation step and is dynamic, it prevents you from taking advantage of [`torch.compile`](./perf_torch_compile), a powerful optimization tool that fuses PyTorch code into fast and optimized kernels.

The *static kv-cache* solves this issue by pre-allocating the kv-cache size to a maximum value which allows you to combine it with torch.compile for up to a 4x speed up.
The *static kv-cache* solves this issue by pre-allocating the kv-cache size to a maximum value which allows you to combine it with `torch.compile` for up to a 4x speed up. Your speed up may vary depending on the model size (larger models have a smaller speed up) and hardware.

> [!WARNING]
> Currently, only [Llama](./model_doc/llama2) and a few other models support static kv-cache and torch.compile. Check [this issue](https://github.com/huggingface/transformers/issues/28981) for a live model compatibility list.
> Currently, only [Llama](./model_doc/llama2) and a few other models support static kv-cache and `torch.compile`. Check [this issue](https://github.com/huggingface/transformers/issues/28981) for a live model compatibility list.
For this example, let's load the [Gemma](https://hf.co/google/gemma-2b) model.
There are three flavors of static kv-cache usage, depending on the complexity of your task:
1. Basic usage: simply set a flag in `generation_config` (recommended);
2. Advanced usage: handle a cache object for multi-turn generation or a custom generation loop;
3. Advanced usage: compile the entire `generate` function into a single graph, if having a single graph is relevant for you.

Select the correct tab below for further instructions on each of these flavors.

> [!TIP]
> Regardless of the strategy used with `torch.compile`, you can avoid shape-related recompilations if you left-pad your LLM inputs to a limited set of values. The [`pad_to_multiple_of` tokenizer flag](https://huggingface.co/docs/transformers/main_classes/tokenizer#transformers.PreTrainedTokenizer.__call__.pad_to_multiple_of) is your friend!
<hfoptions id="static-kv">
<hfoption id="basic usage: generation_config">

For this example, let's use the [Gemma](https://hf.co/google/gemma-2b) model. All we need to do is to:
1. Access the model's `generation_config` attribute and set the `cache_implementation` to "static";
2. Call `torch.compile` on the model to compile the forward pass with the static kv-cache.

And that's it!

```py
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
import os
os.environ["TOKENIZERS_PARALLELISM"] = "false" # To prevent long warnings :)

tokenizer = AutoTokenizer.from_pretrained("google/gemma-2b")
model = AutoModelForCausalLM.from_pretrained(
"google/gemma-2b", device_map="auto"
)
model = AutoModelForCausalLM.from_pretrained("google/gemma-2b", device_map="auto")

model.generation_config.cache_implementation = "static"

model.forward = torch.compile(model.forward, mode="reduce-overhead", fullgraph=True)
input_text = "The theory of special relativity states "
input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")

outputs = model.generate(**input_ids)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
['The theory of special relativity states 1. The speed of light is constant in all inertial reference']
```

There are two ways you can configure the model to use a static kv-cache. For a 7B model on an A100, both methods get a 4x speed up in the forward pass. Your speed up may vary depending on the model size (larger models have a smaller speed up) and hardware. If you're using the [`~GenerationMixin.generate`] method, the speed up is ~3x. The forward pass (which still gets 4x speed up) is only a part of the whole [`~GenerationMixin.generate`] code.
Under the hood, `generate` will attempt to reuse the same cache object, removing the need for re-compilation at each call. Avoiding re-compilation is critical to get the most out of `torch.compile`, and you should be aware of the following:
1. If the batch size changes or the maximum output length increases between calls, the cache will have to be reinitialized, triggering a new compilation;
2. The first couple of calls of the compiled function are slower, as the function is being compiled.

<hfoptions id="static-kv">
<hfoption id="generation_config">
> [!WARNING]
> For a more advanced usage of the static cache, such as multi-turn conversations, we recommend instantiating and manipulating the cache object outside [`~GenerationMixin.generate`]. See the advanced usage tab.
</hfoption>
<hfoption id="advanced usage: control Static Cache">

Access the model's `generation_config` attribute and set the `cache_implementation` to "static".
A [`StaticCache`] object can be passed to the model's [`~GenerationMixin.generate`] under the `past_key_values` argument. The object will retain the cache contents, so you can pass it to a new [`~GenerationMixin.generate`] call to continue generation, like you would do with a dynamic cache.

```py
model.generation_config.cache_implementation = "static"
```
from transformers import AutoTokenizer, AutoModelForCausalLM, StaticCache
import torch
import os
os.environ["TOKENIZERS_PARALLELISM"] = "false" # To prevent long warnings :)

Call torch.compile on the model to compile the forward pass with the static kv-cache.
tokenizer = AutoTokenizer.from_pretrained("google/gemma-2b")
model = AutoModelForCausalLM.from_pretrained("google/gemma-2b", device_map="auto")

```py
compiled_model = torch.compile(model, mode="reduce-overhead", fullgraph=True)
model.forward = torch.compile(model.forward, mode="reduce-overhead", fullgraph=True)
input_text = "The theory of special relativity states "
input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")
prompt_length = input_ids.input_ids.shape[1]
model.generation_config.max_new_tokens = 16

past_key_values = StaticCache(
config=model.config,
max_batch_size=1,
# If you plan to reuse the cache, make sure the cache length is large enough for all cases
max_cache_len=prompt_length+(model.generation_config.max_new_tokens*2),
device=model.device,
dtype=model.dtype
)
outputs = model.generate(**input_ids, past_key_values=past_key_values)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
['The theory of special relativity states 1. The speed of light is constant in all inertial reference frames. 2']

outputs = compiled_model.generate(**input_ids)
tokenizer.batch_decode(outputs, skip_special_tokens=True)
['The theory of special relativity states 1. The speed of light is constant in all inertial reference']
# pass in the generated text and the same cache object to continue generation from where it left off. Optionally, in a
# multi-turn conversation, append the new user input to the generated text.
new_input_ids = outputs
outputs = model.generate(new_input_ids, past_key_values=past_key_values)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
['The theory of special relativity states 1. The speed of light is constant in all inertial reference frames. 2. The speed of light is constant in all inertial reference frames. 3.']
```

Under the hood, `generate` will attempt to reuse the same cache object, removing the need for re-compilation at each call. However, if the batch size or the maximum output length increase between calls, the cache will have to be reinitialized, triggering a new compilation.

</hfoption>
<hfoption id="Static Cache">
> [!TIP]
> If you want to reuse the same [`StaticCache`] object on a new prompt, be sure to reset its contents with the `.reset()` method between calls
A [`StaticCache`] object can be passed to the model's forward pass under the `past_key_values` argument, enabling the use of this object as a static kv-cache. Using this strategy, you can write your own function to decode the next token given the current token and position and cache position of previously generated tokens. You can also pass the [`StaticCache`] object to [`~GenerationMixin.generate`] and use it across calls, like you would do with a dynamic cache.
If you want to go further down a level, the [`StaticCache`] object can also be passed to the model's forward pass under the same `past_key_values` argument. Using this strategy, you can write your own function to decode the next token given the current token and position and cache position of previously generated tokens.

```py
from transformers import LlamaTokenizer, LlamaForCausalLM, StaticCache, logging
Expand Down Expand Up @@ -102,12 +152,9 @@ def decode_one_tokens(model, cur_token, input_pos, cache_position, past_key_valu
return new_token
```

There are a few important things you must do to enable static kv-cache and torch.compile with the `StaticCache` method:

There are a few important things you must do to enable static kv-cache and `torch.compile` with the `StaticCache` method:
1. Initialize the [`StaticCache`] instance before using the model for inference. There you can configure parameters like the maximum batch size and sequence length.

2. Call torch.compile on the model to compile the forward pass with the static kv-cache.

2. Call `torch.compile` on the model to compile the forward pass with the static kv-cache.
3. Set `enable_math=True` in the [torch.backends.cuda.sdp_kernel](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html) context manager to enable the native PyTorch C++ implementation of scaled dot product attention to speed up inference even more.

```py
Expand Down Expand Up @@ -142,8 +189,34 @@ text
'My favorite all time favorite condiment is ketchup. I love it on everything. I love it on my eggs, my fries, my chicken, my burgers, my hot dogs, my sandwiches, my salads, my p']
```

> [!TIP]
> If you want to reuse the [`StaticCache`] object on a new prompt, be sure to reset its contents with the `.reset()` method
</hfoption>
<hfoption id="advanced usage: end-to-end generate compilation">

Compiling the entire `generate` function, in terms of code, is even simpler than in the basic usage: call `torch.compile` on `generate` to compile the entire function. No need to specify the use of the static cache: although it is compatible, dynamic cache (default) was faster in our benchmarks.

```py
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
import os
os.environ["TOKENIZERS_PARALLELISM"] = "false" # To prevent long warnings :)

tokenizer = AutoTokenizer.from_pretrained("google/gemma-2b")
model = AutoModelForCausalLM.from_pretrained("google/gemma-2b", device_map="auto")

model.generate = torch.compile(model.generate, mode="reduce-overhead", fullgraph=True)
input_text = "The theory of special relativity states "
input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")

outputs = model.generate(**input_ids)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
['The theory of special relativity states 1. The speed of light is constant in all inertial reference']
```

As a result, we compile not only the model forward pass, but also all input preparation, logit processor operations, and so on. The result should be a slightly `generate` call, compared to the basic usage example, and the compiled graph may be better suited to more exotic hardware devices or use cases. However, there are severe drawbacks in using this approach:
1. Compilation is much slower;
2. All parameterization of `generate` must be done through `generation_config`;
3. Many warnings and exceptions are suppressed -- we suggest testing with its uncompiled form first;
4. Although we are working on it, it is heavily feature restricted (for instance, at the time of writing, generation does not stop if an EOS token is selected).

</hfoption>
</hfoptions>
Expand Down
11 changes: 6 additions & 5 deletions src/transformers/cache_utils.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
from packaging import version

from .configuration_utils import PretrainedConfig
from .utils import is_hqq_available, is_quanto_available, logging
from .utils import is_hqq_available, is_quanto_available, is_torchdynamo_compiling, logging


if is_quanto_available():
Expand Down Expand Up @@ -398,7 +398,6 @@ def from_legacy_cache(cls, past_key_values: Optional[Tuple[Tuple[torch.FloatTens
def crop(self, max_length: int):
"""Crop the past key values up to a new `max_length` in terms of tokens. `max_length` can also be
negative to remove `max_length` tokens. This is used in assisted decoding and contrastive search."""

# In case it is negative
if max_length < 0:
max_length = self.get_seq_length() - abs(max_length)
Expand Down Expand Up @@ -821,11 +820,13 @@ def __init__(self, config: PretrainedConfig, max_batch_size: int, max_cache_len:
cache_shape = (max_batch_size, self.num_key_value_heads, self.max_cache_len, self.head_dim)
for _ in range(config.num_hidden_layers):
# Note: `mark_static_address` is used to tag the cache as an fixed data pointer, preventing cuda graph
# breaks when updating the cache.
# breaks when updating the cache. It can't be used if the cache code is being compiled (but in that case
# it is not needed anyway)
new_layer_key_cache = torch.zeros(cache_shape, dtype=self.dtype, device=device)
new_layer_value_cache = torch.zeros(cache_shape, dtype=self.dtype, device=device)
torch._dynamo.mark_static_address(new_layer_key_cache)
torch._dynamo.mark_static_address(new_layer_value_cache)
if not is_torchdynamo_compiling():
torch._dynamo.mark_static_address(new_layer_key_cache)
torch._dynamo.mark_static_address(new_layer_value_cache)
self.key_cache.append(new_layer_key_cache)
self.value_cache.append(new_layer_value_cache)

Expand Down
Loading

0 comments on commit 7ffe25f

Please sign in to comment.