Thank you for your excellent work. I have encountered some confusion while trying to reproduce the results:
In snapkv_utils.py, the following snippet is used to select indices and gather corresponding values:
snapkv_utils.py#L62-L65
indices = attn_cache.topk(self.max_capacity_prompt - self.window_size, dim=-1).indices
indices = indices.unsqueeze(-1).expand(-1, -1, -1, head_dim)
k_past_compress = key_states[:, :, :-self.window_size, :].gather(dim = 2, index = indices)
v_past_compress = value_states[:, :, :-self.window_size, :].gather(dim = 2, index = indices)
The indices here are selected based on the scores in attn_cache, which are sorted by their values, rather than by their original positional order.
My concern is whether this approach could lead to incorrect ordering in the KV cache when performing the gather operation for k_past_compress and v_past_compress. Specifically, could this potentially disrupt the alignment between keys and values during subsequent inference steps?
I would appreciate any insights or clarifications on this matter. Thank you!
Thank you for your excellent work. I have encountered some confusion while trying to reproduce the results:
In
snapkv_utils.py,the following snippet is used to select indices and gather corresponding values:snapkv_utils.py#L62-L65
The indices here are selected based on the scores in
attn_cache, which are sorted by their values, rather than by their original positional order.My concern is whether this approach could lead to incorrect ordering in the KV cache when performing the gather operation for
k_past_compressandv_past_compress. Specifically, could this potentially disrupt the alignment between keys and values during subsequent inference steps?I would appreciate any insights or clarifications on this matter. Thank you!