zhangyiss/mlx-examples
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx-examples.git synced 2025-09-01 12:49:50 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

More cache improvements (#1015)

Browse Source 
* fix rotating kv cache for chat use case

* reorg + fixes to caching, unify prompt caching across types and use cases for e.g. caching during a chat

* nit in chat

* fix tests

* fix tests

* fix tests

* docs

* chat command

* comments + docs

* Define meta_state on all Cache implementations

* fixes + trim_prompt_cache api

* fix default model

---------

Co-authored-by: Angelos Katharopoulos <a_katharopoulos@apple.com>
This commit is contained in:
Awni Hannun
2024-10-07 20:45:51 -07:00
committed by GitHub
parent 9bc53fc210
commit fca087be49
43 changed files with 1151 additions and 691 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
llms/mlx_lm/models/nemotron.py
View File

@@ -2,12 +2,12 @@
from dataclasses import dataclass
from functools import partial
from typing import Dict, Optional, Union
from typing import Any, Dict, Optional, Union
import mlx.core as mx
import mlx.nn as nn
from .base import BaseModelArgs, KVCache, create_attention_mask
from .base import BaseModelArgs, create_attention_mask
@dataclass
@@ -94,7 +94,7 @@ class Attention(nn.Module):
self,
x: mx.array,
mask: Optional[mx.array] = None,
cache: Optional[KVCache] = None,
cache: Optional[Any] = None,
) -> mx.array:
B, L, _ = x.shape
@@ -151,7 +151,7 @@ class TransformerBlock(nn.Module):
self,
x: mx.array,
mask: Optional[mx.array] = None,
cache: Optional[KVCache] = None,
cache: Optional[Any] = None,
) -> mx.array:
r = self.self_attn(self.input_layernorm(x), mask, cache)
h = x + r
@@ -215,13 +215,3 @@ class Model(nn.Module):
@property
def layers(self):
return self.model.layers
@property
def head_dim(self):
return (
self.args.head_dim or self.args.hidden_size // self.args.num_attention_heads
)
@property
def n_kv_heads(self):
return self.args.num_key_value_heads