branch: externals/elisa
commit 1615d28474c0fb9328824e26414db145c69c0553
Author: Sergey Kostyaev <kostyaev.serg...@wb.ru>
Commit: Sergey Kostyaev <kostyaev.serg...@wb.ru>
Update documentation
Updated the embedding model to "snowflake-arctic-embed2" in various
configuration sections. Added a new command
`elisa-async-recalculate-embeddings` to recalculate embeddings
asynchronously. Also, added a new configuration option
`elisa-batch-size` for controlling batch size during batch embeddings
calculation.
---
README.org | 23 ++++++++++++++++++-----
1 file changed, 18 insertions(+), 5 deletions(-)
diff --git a/README.org b/README.org
index 3fc9d4003a..29af50a377 100644
--- a/README.org
+++ b/README.org
@@ -88,7 +88,7 @@ I prefer this models:
#+begin_src shell
ollama pull gemma2:9b-instruct-q6_K
ollama pull qwen2.5:3b
- ollama pull chatfire/bge-m3:q8_0
+ ollama pull snowflake-arctic-embed2
#+end_src
**** Complex documents
@@ -203,6 +203,12 @@ Disable all collections.
Removes collection and all its data from index.
+*** elisa-async-recalculate-embeddings
+
+Recalculate embeddings asynchronously. Use it after changing
+~elisa-embeddings-provider~ variable. Can take some time. Works faster
+with ~elisa-batch-embeddings-enabled~.
+
** Configuration
Example configuration.
@@ -222,7 +228,7 @@ Example configuration.
;; (setopt elisa-chat-provider
;; (make-llm-ollama
;; :chat-model "gemma2:9b-instruct-q6_K"
- ;; :embedding-model "chatfire/bge-m3:q8_0"
+ ;; :embedding-model "snowflake-arctic-embed2"
;; ;; set context window to 8k
;; :default-chat-non-standard-params '(("num_ctx" . 8192))))
;;
@@ -230,11 +236,13 @@ Example configuration.
(setopt elisa-chat-provider
(make-llm-ollama
:chat-model "qwen2.5:3b"
- :embedding-model "chatfire/bge-m3:q8_0"
+ :embedding-model "snowflake-arctic-embed2"
:default-chat-temperature 0.1
:default-chat-non-standard-params '(("num_ctx" . 32768))))
;; this embedding model has stong multilingual capabilities
- (setopt elisa-embeddings-provider (make-llm-ollama :embedding-model
"chatfire/bge-m3:q8_0"))
+ (setopt elisa-embeddings-provider (make-llm-ollama :embedding-model
"snowflake-arctic-embed2"))
+ ;; enable batch embeddings for faster processing
+ (setopt elisa-batch-embeddings-enabled t)
:config
;; searxng works better than duckduckgo in my tests
(setopt elisa-web-search-function 'elisa-search-searxng))
@@ -293,7 +301,7 @@ Example configuration.
+ ~elisa-semantic-split-function~:
* Type: Function
- * Description: Function used to split text into semantically meaningful
chunks.
+ * Description: Function used to split text into semantically meaningful
chunks.
* Default: ~elisa-split-by-paragraph~
+ ~elisa-prompt-rewriting-enabled~:
@@ -313,6 +321,11 @@ Example configuration.
* Type: Boolean
* Description: Enable batch embeddings if supported.
++ ~elisa-batch-size~:
+ * Type: Integer
+ * Description: Batch size to send to provider during batch embeddings
calculation.
+ * Default: 300
+
**** Web Search and Integration
+ ~elisa-searxng-url~:
