fix readme

Signed-off-by: Dominic789654 <xliu29@gmu.edu>

Author: Dominic789654

README.md

@@ -77,7 +77,7 @@ Finally we provide wrapper presses that can be combined with other presses:
 - `PerLayerCompressionPress` ([source](kvpress/presses/per_layer_compression_press.py)): compress each layer with a different compression ratio (experimental)
 - `ComposedPress` ([source](kvpress/presses/composed_press.py)): compose multiple presses together by chaining their forward hooks
 - `KeyRerotationPress` ([source](kvpress/presses/key_rerotation_press.py)): rerotate pruned keys to have continuous RoPE embeddings
-- `ChunkKVPress` ([source](kvpress/presses/chunkkv_press.py), [paper](https://arxiv.org/abs/2502.00299)): implements the ChunkKV compression method that selects whole chunks based on their importance scores. This approach differs from ChunkPress by maintaining chunk-level granularity during selection, which helps preserve local attention patterns. The method is particularly effective for long sequences where maintaining contextual coherence is important.
+- `ChunkKVPress` ([source](kvpress/presses/chunkkv_press.py), [paper](https://arxiv.org/abs/2502.00299)): compresses by selecting important chunks, preserving semantic coherence
 - `ChunkPress` ([source](kvpress/presses/chunk_press.py), [paper](https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00716/125280)): compress the KV cache on each sequence chunk separately. This can yield to more uniform compression across long sequences
 - `CriticalKVPress` and `CriticalAdaKVPress` ([source](kvpress/presses/criticalkv_press.py), [paper](https://arxiv.org/abs/2502.03805)): refine the scores using the L1 norm of Wo @ values, coupled with a two-stage selection.
 

kvpress/presses/chunk_press.py

@@ -50,7 +50,6 @@ def compress(
         assert attentions is None, "ChunkPress does not support attentions."
 
         kv_len = keys.shape[2]
-
         indices = []
         for i in range(0, kv_len, self.chunk_length):
             chunk_scores = self.press.score(