Skip to content

Update for new version of HF transformers. #104

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
manueldeprada wants to merge 3 commits into from
Closed

Update for new version of HF transformers. #104

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
09fbcbb
Update base_press.py
manueldeprada Jul 23, 2025
1738a02
Update test_pipeline.py
manueldeprada Jul 23, 2025
9986c31
Update attention_patch.py
manueldeprada Aug 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions kvpress/attention_patch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,6 +78,7 @@ def wrapper(module, query, key, value, attention_mask, dropout, **kwargs):
batch_indices, head_indices, seq_indices = module.masked_key_indices
key[batch_indices, head_indices, seq_indices] = k[batch_indices, head_indices]

kwargs["cu_seq_lens_k"][-1] = key.shape[-2]
return func(module, query, key, value, attention_mask, dropout, **kwargs)

return wrapper
Expand Down
22 changes: 11 additions & 11 deletions kvpress/presses/base_press.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,23 +131,23 @@ def forward_hook(self, module: nn.Module, input: list[torch.Tensor], kwargs: dic
return output

if isinstance(cache, QuantizedCache):
keys = cache._dequantize(cache._quantized_key_cache[module.layer_idx])
values = cache._dequantize(cache._quantized_value_cache[module.layer_idx])
keys = cache.cache_processor._dequantize(cache.cache_processor._quantized_keys[module.layer_idx])
values = cache.cache_processor._dequantize(cache.cache_processor._quantized_values[module.layer_idx])
else:
keys = cache.key_cache[module.layer_idx]
values = cache.value_cache[module.layer_idx]
keys = cache.layers[module.layer_idx].keys
values = cache.layers[module.layer_idx].values

keys, values = self.compress(module, hidden_states, keys, values, output[1], kwargs)

if isinstance(cache, QuantizedCache):
cache._quantized_key_cache[module.layer_idx] = cache._quantize(keys, axis=cache.axis_key)
cache._quantized_value_cache[module.layer_idx] = cache._quantize(values, axis=cache.axis_value)
cache.key_cache[module.layer_idx] = torch.zeros(0, dtype=keys.dtype, device=keys.device)
cache.value_cache[module.layer_idx] = torch.zeros(0, dtype=keys.dtype, device=keys.device)
cache._seen_tokens = keys.shape[2]
cache.cache_processor._quantized_keys[module.layer_idx] = cache.cache_processor._quantize(keys, axis=cache.cache_processor.axis_key)
cache.cache_processor._quantized_values[module.layer_idx] = cache.cache_processor._quantize(values, axis=cache.cache_processor.axis_value)
cache.layers[module.layer_idx].keys = torch.zeros(0, dtype=keys.dtype, device=keys.device)
cache.layers[module.layer_idx].values = torch.zeros(0, dtype=keys.dtype, device=keys.device)
cache.cache_processor.erased_length = keys.shape[2]
else:
cache.key_cache[module.layer_idx] = keys
cache.value_cache[module.layer_idx] = values
cache.layers[module.layer_idx].keys = keys
cache.layers[module.layer_idx].values = values

return output

Expand Down
5 changes: 2 additions & 3 deletions tests/test_pipeline.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@

import pytest
import torch
from transformers import AutoTokenizer, DynamicCache, QuantizedCacheConfig, QuantoQuantizedCache
from transformers import AutoTokenizer, DynamicCache, QuantoQuantizedCache
from transformers.utils import is_optimum_quanto_available

from kvpress import ExpectedAttentionPress
Expand Down Expand Up @@ -82,8 +82,7 @@ def test_pipeline_with_quantized_cache(kv_press_danube_pipeline, caplog): # noq
context = "This is a test article. It was written on 2022-01-01."
questions = ["When was this article written?"]
press = ExpectedAttentionPress(compression_ratio=0.4)
config = QuantizedCacheConfig(nbits=4)
cache = QuantoQuantizedCache(config)
cache = QuantoQuantizedCache(nbits=4)
answers = kv_press_danube_pipeline(context, questions=questions, press=press, cache=cache)["answers"]

assert len(answers) == 1
Expand Down