char-fork/llama.cpp

git clone https://git.t4t.associates/char-fork/llama.cpp

Aleksander Grygierui: Stores split refactor (#27240)521a64cd0

char/aliceai-t5-moe
brancheschar/aliceai-t5-moeall refs →
924 B34 linesraw
1/**
2 * Cache configuration constants
3 */
4
5/**
6 * Default cache limits when no per-cache overrides are given.
7 */
8export const CACHE = {
9	/** Default maximum number of entries in a cache */
10	DEFAULT_MAX_ENTRIES: 100,
11	/** Default TTL (Time-To-Live) for cache entries in milliseconds (5 minutes) */
12	DEFAULT_TTL_MS: 5 * 60 * 1000
13} as const;
14
15/**
16 * TTL and size for the model props cache.
17 * Props don't change frequently, so we can cache them longer.
18 */
19export const MODEL_PROPS_CACHE = {
20	/** Maximum number of model props to cache */
21	MAX_ENTRIES: 50,
22	/** TTL for model props cache entries in milliseconds (10 minutes) */
23	TTL_MS: 10 * 60 * 1000
24} as const;
25
26/**
27 * TTL and size for the MCP resource cache.
28 */
29export const MCP_RESOURCE_CACHE = {
30	/** Maximum number of MCP resources to cache */
31	MAX_ENTRIES: 50,
32	/** TTL for MCP resource cache entries in milliseconds (5 minutes) */
33	TTL_MS: 5 * 60 * 1000
34} as const;