241b14932b
updated README with latest information
2025-12-08 00:56:12 +01:00
ec8dd6e8a0
added max context size configuration option
2025-12-08 00:55:58 +01:00
b3ad72a7a2
updated default model used to ministral 8b
...
minstral ai just released their latest generation of open source models.
Ministral 8b has great performance even with very small quantizations.
So for now the ministral 8b q2 will be used as the new default. This
significantly dereases the size of the container while improving performance
2025-12-08 00:41:12 +01:00
642ffc60c6
update llama-cpp bindings
...
there are multiple new versions of llama-cpp. even the rust bindings are
not up to date which can cause issues when using bleeding edge models or quantizations
updatey my fork of llama-cpp-rs
2025-12-08 00:41:09 +01:00
44bb6244f3
allow server host and port configuration
2025-12-07 18:13:24 +01:00
40e49eb508
update config to include host and port confiugration
2025-12-07 18:13:03 +01:00
0831dfad1c
rename reasoning field for decision questions
...
Rust / build (push) Has been cancelled
by removing summary the model is less likely to try to summarize the
document and using only as a keyword reinforces keeping the reasoning
short.
2025-11-24 02:18:11 +01:00
fd5bf7c5ed
fix default num gpu layers
...
i thought zero would mean unlimited but that seems to be incorrect.
Somewhere I found a reference to `-1` as unlimited by llama-cpp-rs
bindings only allow unsigned numbers. Maybe I sould investigate this
further in the future
2025-11-24 02:08:14 +01:00
cecd2941bc
chore: Release paperless-llm-workflows version 0.3.1
2025-11-23 13:16:22 +01:00
ee253139c9
update gitignore
2025-11-23 13:14:04 +01:00
fe9cdc391a
configure cargo release for this project
2025-11-23 13:12:21 +01:00
66a7850e28
update changelog for next release
2025-11-23 13:07:39 +01:00
b89aa8f65f
chore: update openapi.json to include decision
2025-11-23 13:05:31 +01:00
5176b29587
chore: code readability implement clippy suggestions
2025-11-23 13:04:23 +01:00
96bd2c9ce0
Merge branch 'feature/workflow-decision-tag-flow'
Rust / build (push) Has been cancelled
2025-11-22 19:57:06 +01:00
9b7ff13bbd
add decision feature to README
2025-11-22 19:57:04 +01:00
183ec86ea4
update CHANGELOG
2025-11-22 19:57:02 +01:00
428f34c980
fix all compiler warnings
2025-11-22 19:57:00 +01:00
65f65f6cb7
implement decision handler and document tag updates
2025-11-22 19:56:57 +01:00
a94f75dc54
add type to guide the model through the decision process
2025-11-22 19:56:56 +01:00
4627591576
implement decision webhook receive and add request to processing queue
2025-11-22 19:56:54 +01:00
2e90af0e98
change return type of extract function
...
extract can take any schema, but returns a FieldExtract which is not as
flexible as the grammar. Given that more complex queries may not be
mapped to FieldExtract. So changing the return type to Value allows more
flexible reusage of `extract` and puts the burdon on the caller to
convert value to the desired type
2025-11-22 19:56:52 +01:00
1b15ec9f26
add skeleton code to support new llm workflow kind
2025-11-22 19:56:49 +01:00
5ecf8998b5
Create FUNDING.yml
2025-11-22 16:11:14 +01:00
aadf19774c
fix docker build file after rename of project
Rust / build (push) Waiting to run
2025-11-22 15:29:40 +01:00
df13239df5
release 0.3.0 api workflow integration version
Rust / build (push) Waiting to run
2025-11-22 15:18:20 +01:00
67271241d0
update container names and references in README
2025-11-22 15:16:32 +01:00
ef1a0f380b
add sequence diagram to README for better overview of process
Rust / build (push) Waiting to run
2025-11-22 15:12:51 +01:00
d1bbc73478
add link to online swagger docs
Rust / build (push) Waiting to run
2025-11-22 14:44:46 +01:00
0b3a269b10
checkin current api specification to repository
Rust / build (push) Waiting to run
2025-11-22 14:11:31 +01:00
392d48a356
add command line flag to easily dump the api spec
2025-11-22 14:11:11 +01:00
08986aa837
move documentation for api endpoints to swagger
2025-11-22 14:10:45 +01:00
745434e642
restructure README
Rust / build (push) Waiting to run
2025-11-22 13:49:25 +01:00
249e65c55f
rename the project to paperless-llm-workflows
...
given the server architecture and new integration into paperless
workflows a new name describes better what this project is about
2025-11-22 02:17:56 +01:00
c5cb0aaf22
Merge branch 'feature/defer-document-update'
Rust / build (push) Waiting to run
2025-11-22 01:13:10 +01:00
1ad1ffb543
update Changelog
2025-11-22 01:12:28 +01:00
40013b5e48
fix all compiler warnings
2025-11-22 01:10:22 +01:00
7cfe55bcdf
async doc updates to paperless instance
2025-11-20 22:27:12 +01:00
31726c3f01
Merge branch 'feature/server-api-spec'
Rust / build (push) Has been cancelled
2025-11-17 15:16:00 +01:00
165ddaa606
improve model load and unload behavior
2025-11-17 15:15:09 +01:00
91d74b2f2d
add some mote doc string to important types
...
types exposed via the api get some more docstring to include more
information in the api specification
2025-11-17 01:43:03 +01:00
3e5d636687
add CHANGELOG entry about openapi support
2025-11-17 01:43:01 +01:00
715853511a
setup openapi specs and swagger ui
...
add interactive api documentation to the server to enable browsable
endpoint docs
2025-11-17 01:42:59 +01:00
410b25628e
reorder code for better grouping
2025-11-17 01:42:58 +01:00
3e572167c5
add utoipa for documenting hook endpoints
2025-11-17 01:42:55 +01:00
fe10c74b00
make model loading fallible
2025-11-17 01:42:29 +01:00
ed2f34637e
Merge branch 'feature/next-tag-control'
Rust / build (push) Waiting to run
2025-11-17 00:05:50 +01:00
a9a9753864
add CHANGELOG entry
2025-11-17 00:04:59 +01:00
bda34530a5
add next_tag parameter to webhook parameters
...
allow webhook to sepecify the next tag that will be applied to the
document after processing has finished. This allows for creating more
complex workflows using tags to setup processing stages
2025-11-17 00:02:24 +01:00
9f4e7a4340
chore: fix all warnings
2025-11-16 22:15:05 +01:00
