Skip to content

Commit 873a4bb

Browse files
committed
SimpleChatTC:SettingsDefault:Enable cache prompt api option
1 parent fd372a6 commit 873a4bb

File tree

2 files changed

+15
-6
lines changed

2 files changed

+15
-6
lines changed

tools/server/public_simplechat/readme.md

Lines changed: 14 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -583,21 +583,30 @@ Initial go at a pdftext tool call. It allows web / local pdf files to be read an
583583
extracted and passed to ai model for further processing, as decided by ai and end user. One could
584584
either work with the full pdf or a subset of adjacent pages.
585585

586-
SimpleProxy
586+
SimpleProxy updates
587587
* Convert from a single monolithic file into a collection of modules.
588588
* UrlValidator to cross check scheme and domain of requested urls,
589589
the whitelist inturn picked from config json
590590
* Helpers to fetch file from local file system or the web, transparently
591+
* Help check for needed modules before a particular service path is acknowledged as available
592+
through /aum service path
591593

592-
Chances are for ai models which dont support tool calling, things will be such that the tool calls
593-
meta data shared will be silently ignored without much issue. So enabling tool calling feature by default,
594-
so that in case one is using a ai model with tool calling the feature is readily available for use.
594+
Settings/Config default changes
595595

596-
Revert SlidingWindow ChatHistory in Context from last 10 to last 5 (rather 2 more then origianl,
596+
* Chances are for ai models which dont support tool calling, things will be such that the tool calls
597+
meta data shared will be silently ignored without much issue. So enabling tool calling feature by
598+
default, so that in case one is using a ai model with tool calling the feature is readily available
599+
for use.
600+
601+
* Revert SlidingWindow ChatHistory in Context from last 10 to last 5 (rather 2 more then origianl,
597602
given more context support in todays models) by default, given that now tool handshakes go through
598603
the tools related side channel in the http handshake and arent morphed into normal user-assistant
599604
channel of the handshake.
600605

606+
* Enable CachePrompt api option given that tool calling based interactions could involve chat sessions
607+
having ai responses built over multiple steps of tool callings etal. So independent of our client side
608+
sliding window based drop off or even before they kick in, this can help in many cases.
609+
601610

602611
#### ToDo
603612

tools/server/public_simplechat/simplechat.js

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1356,7 +1356,7 @@ class Me {
13561356
"temperature": 0.7,
13571357
"max_tokens": 2048,
13581358
"n_predict": 2048,
1359-
"cache_prompt": false,
1359+
"cache_prompt": true,
13601360
//"frequency_penalty": 1.2,
13611361
//"presence_penalty": 1.2,
13621362
};

0 commit comments

Comments
 (0)