chungyau97
|
342adff8cb
|
fix weird error
|
2024-02-16 20:20:17 +08:00 |
Henry
|
778e024c02
|
update query engine tool
|
2024-02-14 18:40:35 +08:00 |
Ilango
|
4e8bf4903d
|
Merge pull request #1687 from 0xi4o/bug/scrap-limit
Fix: relative links method and limit not applying to manage links
|
2024-02-12 12:35:40 +05:30 |
|
Ilango
|
5471a4c9aa
|
Show error when relative links method is not set and allow 0 as limit value
|
2024-02-12 12:01:19 +05:30 |
Henry Heng
|
2ef0f813c6
|
Merge pull request #1703 from FlowiseAI/feature/Retriever-Output
Feature/Add document json output
|
2024-02-10 22:12:23 +08:00 |
|
Henry Heng
|
55dad31438
|
Merge pull request #1677 from fletch-ai/bugfix/Postgres-SSL-Fix
Fixing Postgres so that similaritySearchVectorWithScore actually connects via TLS/SSL when specified
|
2024-02-10 22:10:36 +08:00 |
|
Henry
|
caf54bf31b
|
add document json output
|
2024-02-09 16:07:34 +08:00 |
|
Henry Heng
|
1f2e58c5d5
|
Merge pull request #1697 from Jaredude/bugfix/dynamob-memory-store
DynamoDB Chat Memory fix
|
2024-02-09 01:36:38 +08:00 |
Darien Kindlund
|
907d5c7ef7
|
Bumping version since this will change might break legacy Postgres chatflows that used TLS/SSL. This will help users remind them to upgrade their corresponding nodes.
|
2024-02-08 12:09:26 -05:00 |
|
Darien Kindlund
|
4a75396325
|
So apparently TypeORMVectorStore requires the field 'username' while Pool expects it to be 'user'. Need to populate both to avoid error messages.
|
2024-02-08 12:09:26 -05:00 |
|
Darien Kindlund
|
9af4eaaa8c
|
Fixing linting issues.
|
2024-02-08 12:09:26 -05:00 |
|
Darien Kindlund
|
e110a49a32
|
Migrating all TLS/SSL settings into additionalConfig so that there is consistency in how every function connects with Postgres in the same manner.
|
2024-02-08 12:09:26 -05:00 |
Jared Tracy
|
dd89af8a09
|
refactor (this as any) usage
|
2024-02-08 09:04:46 -06:00 |
|
Jared Tracy
|
d0b1980482
|
Removes prior comment
|
2024-02-07 22:08:41 -06:00 |
|
Jared Tracy
|
d7f9c07381
|
return tableName, partiionKey, dynamoKey from initalizeDynamoDB
|
2024-02-07 22:07:12 -06:00 |
|
Henry
|
bc054d2fe1
|
add fix for override session id
|
2024-02-08 11:55:30 +08:00 |
|
Jared Tracy
|
3f0f0e4d28
|
DynamoDB Chat Memory fix
Fixes #1624
Please note comment in class BufferMemoryExtended for further discussion, if necessary
|
2024-02-07 20:18:06 -06:00 |
|
Henry Heng
|
ee2bd518c3
|
Merge pull request #1672 from KennyVaneetvelde/feature/improve_structured_output_parser_flexibility
Add a zod-schema based JSON output parser ("Advanced Structured Output Parser")
|
2024-02-07 14:55:46 +08:00 |
Ilyes Tascou
|
19fb13baf0
|
fix for linting
|
2024-02-06 14:36:32 +01:00 |
|
Ilyes Tascou
|
2bb2a7588a
|
add recursive option for folder-loader
|
2024-02-06 14:25:40 +01:00 |
Kenny Vaneetvelde
|
90e6a804e4
|
Make the zod schema a main parameter instead of an additional param
|
2024-02-06 13:02:30 +01:00 |
|
Ilango
|
c2ae7e138c
|
Apply limit to selectedLinks even when relative links method is not specified
|
2024-02-06 14:40:19 +05:30 |
|
Ilyes Tascou
|
011a0a75c3
|
add kotlin files to folder-loader
|
2024-02-05 17:20:05 +01:00 |
Darien Kindlund
|
5da3e3cc3e
|
Adding proper TLS/SSL support to poolOptions so that the similaritySearchVectorWithScore function actually connects to Postgres via TLS/SSL when specified in the additionalConfig.
|
2024-02-04 12:27:18 -05:00 |
|
Kenny Vaneetvelde
|
62f70ab6ef
|
Revert the changes to the original parser & template
|
2024-02-03 21:25:26 +01:00 |
|
Kenny Vaneetvelde
|
95b251f02a
|
Make it a separate node instead
|
2024-02-03 21:22:49 +01:00 |
|
Kenny Vaneetvelde
|
288e451161
|
Improve flexibility of the structured output parser by allowing the user to input an example JSON and automatically converting it to a zod scheme
|
2024-02-03 19:43:44 +01:00 |
|
Henry Heng
|
340e85da91
|
Merge pull request #1611 from FlowiseAI/feature/LlamaIndex
Feature/llama index
|
2024-02-02 20:00:21 +00:00 |
|
Henry Heng
|
37828de664
|
Merge pull request #1659 from FlowiseAI/feature/Vec2Doc
chore/update vec 2 doc node
|
2024-02-01 19:38:27 +00:00 |
Rafael Reis
|
1522acbf5a
|
Add gpt-3.5-turbo-0125 to ChatOpenAI.ts
|
2024-02-01 16:19:30 -03:00 |
|
Henry
|
7881f295ab
|
update vec 2 doc node
|
2024-02-01 19:18:14 +00:00 |
|
Henry
|
5c6b5b233c
|
fix top K
|
2024-02-01 13:14:02 +00:00 |
|
Henry
|
214b312fe5
|
add input moderation to conversation chain
|
2024-01-30 17:57:11 +00:00 |
|
Henry
|
4d6881b506
|
add return source documens to retriever tool
|
2024-01-30 14:12:55 +00:00 |
|
Henry Heng
|
18c9c1cc51
|
Merge pull request #1632 from fletch-ai/bugfix/AzureOpenAIEmbedding-Batch-Size
Bugfix/AzureOpenAIEmbedding - Updated Default Batch Size from 1 to 100
|
2024-01-29 10:12:22 +00:00 |
|
Henry Heng
|
985e454a7f
|
Merge pull request #1631 from fletch-ai/feature/airtable-include-fields
Feature/airtable include fields
|
2024-01-29 10:11:31 +00:00 |
|
Darien Kindlund
|
63665b37ce
|
Switching Redis TTL from EX to PX to match TTL in milliseconds (as specified in the input param).
|
2024-01-28 21:58:04 -08:00 |
|
Darien Kindlund
|
9f4619e408
|
Switching default batch size from 1 to 100 in order to support higher volume of document loading into vector stores before timing out.
|
2024-01-28 21:52:40 -08:00 |
|
Darien Kindlund
|
905c9fc2be
|
Reverting version bump
|
2024-01-28 20:49:21 -08:00 |
|
Darien Kindlund
|
b960f061eb
|
Clarifying that the Limit value is ignored when Return All is set to true.
|
2024-01-28 20:42:19 -08:00 |
|
Darien Kindlund
|
66eef84633
|
Forgot to make maxRecords optional now
|
2024-01-28 20:42:19 -08:00 |
|
Darien Kindlund
|
37945fc998
|
The loadAll() function should ignore any maxRecords specified, because the intention is the load *all* of the records. Also, marking both the Return All and Limit params as optional, so as to not confuse the user. Making them both required adds a lot of confusion that doesn't make sense. Ideally, the user either specifies Return All OR specifies the Limit value but not BOTH. It seems there's no way to define "conditional requirements" in Flowise params, so it's better to make both params optional.
|
2024-01-28 20:42:19 -08:00 |
|
Darien Kindlund
|
dc39d7e2be
|
So Airtable API expects a maxRecords value to be the total set of records you want across multiple API calls. If the maxRecords is greater than 100, then it will provide pagination hints accordingly.
|
2024-01-28 20:42:19 -08:00 |
|
Darien Kindlund
|
9b71f683ff
|
Support pagination even in loadLimit(), so that if a user wants to load more than 100 records but not all of them, they can. Currently, there's a bug where the document loader doesn't work on loading more than 100 records because of internal Airtable API limitations. The Airtable API can only fetch up to 100 records per call - anything more than that requires pagination.
|
2024-01-28 20:42:19 -08:00 |
|
Darien Kindlund
|
2237b1ab16
|
Fix worked, removing debug logging, and bumped node version.
|
2024-01-28 20:42:19 -08:00 |
|
Darien Kindlund
|
3b788e42e1
|
When you switch from GET to POST, you're supposed to adjust the Airtable URL to use the /listRecords endpoint. I didn't RTFM clearly. This is currently documented here: https://support.airtable.com/docs/enforcement-of-url-length-limit-for-web-api-requests
|
2024-01-28 20:42:19 -08:00 |
|
Darien Kindlund
|
456dfabc6f
|
For some reason, Airtable doesn't like the POST operations, so I need to add logging to figure out why this happens
|
2024-01-28 20:42:18 -08:00 |
|
Darien Kindlund
|
8ae848110e
|
Clarifying the description for the optional fields param
|
2024-01-28 20:42:18 -08:00 |
|
Darien Kindlund
|
72ec7878b6
|
Added more error checking and also fixed yet more build errors
|
2024-01-28 20:42:18 -08:00 |
|
Darien Kindlund
|
1a7cb5a010
|
Fixing more build errors
|
2024-01-28 20:42:18 -08:00 |