CTran
a40cdc7b75
Cotran/intent ( #339 )
...
* add else
* integrate hallucination
* remove test
2024-12-08 08:56:35 -08:00
Shuguang Chen
afec644789
Update the logic of intent detection
2024-12-06 14:14:44 -08:00
Shuguang Chen
79eafc0241
Update ArchBaseHandler
2024-12-06 14:07:01 -08:00
Shuguang Chen
2fd8a5a06d
Update Arch-Guard and corresponding e2e test
2024-12-06 13:41:18 -08:00
Shuguang Chen
818ec0a9db
Fix the issue of Arch-Intent
2024-12-06 11:53:46 -08:00
Shuguang Chen
4fcfd83639
Refine model_server
2024-12-05 15:19:41 -08:00
Shuguang Chen
a5bd005411
Fix a bug in function_calling.py
2024-12-05 12:06:56 -08:00
Shuguang Chen
09f7e1e604
Update guardrail_handler and its associated tests
2024-12-05 11:30:58 -08:00
Shuguang Chen
b686cf8b87
refactor model_handler
2024-12-05 11:00:22 -08:00
Shuguang Chen
afe1410b37
Init update on model_server
2024-12-04 16:41:30 -08:00
CTran
cadd3cdaf9
hallucination with log probs ( #281 )
...
* first init
* fix
* fix test
* new implemenetation
* fix bug
* fix bug
* fix bug
* address issue
* address issues
* address comments
* fix test
* fix
* move constatns
* remove consts
2024-11-27 15:17:02 -08:00
Adil Hafeez
0ff3d43008
remove dependency on docker-compose when starting up archgw ( #305 )
2024-11-26 13:13:02 -08:00
Adil Hafeez
d1dd8710a4
release 0.1.2 ( #266 )
2024-11-12 23:56:33 -08:00
Adil Hafeez
a72bb804eb
add support for jaeger tracing ( #229 )
2024-11-07 22:11:00 -06:00
CTran
fb67788be0
add prefill and test ( #236 )
...
* add prefill and test
* fix stream
* fix
* feedback
* address comments
* update
* add e2e test
* fix e2e test
* update fix
* fix
* address cmt
* address cmt
2024-11-07 11:59:29 -08:00
Adil Hafeez
e462e393b1
Use large github action machine to run e2e tests ( #230 )
2024-10-30 17:54:51 -07:00
Adil Hafeez
60299244b9
Improve Gradio UI and fix arch_state bug ( #227 )
2024-10-29 11:27:13 -07:00
José Ulises Niño Rivera
662a840ac5
Add support for streaming and fixes few issues (see description) ( #202 )
2024-10-28 17:05:06 -07:00
CTran
25dddcbfd9
fix model server stop process ( #217 )
...
* fix model server stop process
* replace
* replace
* add test
* add multiple pids test
* add check install for linux
* reformat
2024-10-24 19:21:47 -07:00
Salman Paracha
708fa15a9b
HR agent demo ( #206 )
...
* commiting my hr_agent branch
* updating the HR agent config
* pushing to remote
* fix hr agent
* committing to merge with main
* updating to merge from main
* updating the demo and model-server-tests to pull from poetry
* updating the poetry.lock files
* updating based on feedback
* updated sysmte prompt for hr_agent
---------
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
Co-authored-by: Adil Hafeez <adil@katanemo.com>
2024-10-23 14:32:40 -07:00
Adil Hafeez
dd1c7be706
Pass tool call and app function response back in metadata ( #193 )
2024-10-18 13:25:39 -07:00
CTran
8e54ac20d8
Refactor model server hardware config + add unit tests to load/request to the server ( #189 )
...
* remove mode/hardware
* add test and pre commit hook
* add pytest dependieces
* fix format
* fix lint
* fix precommit
* fix pre commit
* fix pre commit
* fix precommit
* fix precommit
* fix precommit
* fix precommit
* fix precommit
* fix precommit
* fix precommit
* fix precommit
* fix precommit
* fix precommit
2024-10-16 16:58:10 -07:00
Co Tran
b1746b38b4
concatenate history of user messages for hallucination ( #177 )
...
* concatenate history of user messages for hallucination
* add history of messages
* fix gpt to not arch
* add model prefix
* fix
* correct init of user_messages
* fmt
* fix test
2024-10-15 11:43:05 -07:00
Adil Hafeez
7d5f760884
Improve cli ( #179 )
2024-10-10 17:44:41 -07:00
Co Tran
f9e3a052fc
change nli model ( #167 )
...
* change nli model
* Fix bug in hallucination
---------
Co-authored-by: Shuguang Chen <54548843+nehcgs@users.noreply.github.com>
2024-10-09 19:10:08 -07:00
Shuguang Chen
3b7c58698f
Update model_server ( #164 )
...
* Update model server
* Delete model_server/.vscode/settings.json
* Update loader.py
* Fix errors
* Update log mode
2024-10-09 18:04:52 -07:00
Co Tran
8b5db45507
Fix gpu dependency and only leverage onnx when GPU is available ( #157 )
...
* replacing appending instead of write
* fix eetq dependency
* gpu guard required eetq
* fix bug when gpu is available
* fix for gpu device
* reverse
* fix
* replace gpu -> cuda
2024-10-09 11:42:05 -07:00
Co Tran
5c4a6bc8ff
lint + formating with black ( #158 )
...
* lint + formating with black
* add black as pre commit
2024-10-09 11:25:07 -07:00
Salman Paracha
b63a01fe82
Salmanap/fix network agent demo ( #153 )
...
* staging my changes to re-based from main
* adding debug statements to rust
* merged with main
* ready to push network agent
* removed the incomplete sql example
---------
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
2024-10-08 22:19:20 -07:00
Co Tran
e62c6e75ea
fix dependcy + logg info ( #148 )
2024-10-08 16:42:40 -07:00
Co Tran
80d2229053
Cotran/onnx conversion ( #145 )
...
* onnx replacement
* onnx conversion for nli and embedding model
* fix naming
* fix naming
* fix naming
* pin version
2024-10-08 14:37:48 -07:00
Salman Paracha
3ed50e61d2
ensure that we can call the new api.fc.archgw.com url, logging fixes … ( #142 )
...
* ensure that we can call the new api.fc.archgw.com url, logging fixes and minor cli bug fixes
* fixed a bug where model_server printed on terminal after start script stopped running
* updating the logo and fixing the website styles
* updated the branch with feedback from Co and Adil
---------
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
2024-10-08 12:40:24 -07:00
Co Tran
b1fa127704
Hallucination integration with rust ( #122 )
2024-10-07 18:38:55 -07:00
Co Tran
93abe553e3
formating and mointoring change ( #136 )
2024-10-07 15:21:05 -07:00
Adil Hafeez
96686dc606
Serialize tool calls for Arch FC ( #131 )
...
* Serialize tool calls
* fix int tests
2024-10-07 00:03:25 -07:00
Salman Paracha
b60ceb9168
model server build ( #127 )
...
* first commit to have model_server not be dependent on Docker
* making changes to fix the docker-compose file for archgw to set DNS_V4 and minor fixes with the build
* additional fixes for model server to be separated out in the build
* additional fixes for model server to be separated out in the build
* fix to get model_server to be built as a separate python process. TODO: fix the embeddings logs after cli completes
* fixing init to pull tempfile using the tempfile python package
---------
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
2024-10-06 18:21:43 -07:00
Adil Hafeez
2a747df7c0
don't compute embeddings for names and other fixes see description ( #126 )
...
* serialize tools - 2
* fix int tests
* fix int test
* fix unit tests
2024-10-05 19:25:16 -07:00
Salman Paracha
701187474f
load_models checks for device before getting the BGE or NLI model loaded in memory. Was defaulting to CPU. And removed gunk for load_sql ( #119 )
...
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
2024-10-04 13:09:35 -07:00
Co Tran
7d38ef9719
Cotran/hallucination integration ( #115 )
...
* fix fc integration
* fix integration
* remove file
* Update arch_fc.py
* create model server hallucination detection class
2024-10-04 11:05:25 -07:00
Salman Paracha
dc57f119a0
archgw cli ( #117 )
...
* initial commit of the insurange agent demo, with the CLI tool
* committing the cli
* fixed some field descriptions for generate-prompt-targets
* CLI works with buil, up and down commands. Function calling example works stand-alone
* fixed README to install archgw cli
* fixing based on feedback
* fixing based on feedback
---------
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
2024-10-03 18:21:27 -07:00
Adil Hafeez
1b57a49c9d
add support for default target ( #111 )
...
* add support for default target
* add more fixes
2024-10-02 20:43:16 -07:00
Co Tran
ed50d29ccf
fix fc integration ( #110 )
...
* fix fc integration
* fix integration
* remove file
* Update arch_fc.py
2024-10-01 19:20:28 -07:00
Co Tran
17a643c410
ArchFC endpoint integration ( #94 )
...
* integration
* mopdify docker file
* add params and fix python lint
* fix empty context and tool calls
* address comments
* revert port
* fix bug merge
* fix environment
* fix bug
* fix compose
* fix merge
2024-10-01 12:47:26 -07:00
Salman Paracha
8654d3d5c5
simplify developer getting started experience ( #102 )
...
* Fixed build. Now, we have a bare bones version of the docker-compose file with only two services, archgw and archgw-model-server. Tested using CLI
* some pre-commit fixes
* fixed cargo formatting issues
* fixed model server conflict changes
---------
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
2024-10-01 10:02:23 -07:00
Adil Hafeez
f4395d39f9
Fold function_resolver into model_server ( #103 )
2024-10-01 09:13:50 -07:00
José Ulises Niño Rivera
b0ce5eca93
Rename bolt_config to arch_config ( #100 )
...
Signed-off-by: José Ulises Niño Rivera <junr03@users.noreply.github.com>
2024-09-30 18:47:35 -07:00
Adil Hafeez
ea86f73605
rename envoyfilter => arch ( #91 )
...
* rename envoyfilter => arch
* fix more files
* more fixes
* more renames
2024-09-27 16:41:39 -07:00
Co Tran
d5d79256b0
remove guard config json ( #70 )
...
* remove guard config json
* formating
2024-09-24 13:33:31 -07:00
Adil Hafeez
eff4cd9826
improve response handling ( #71 )
2024-09-23 22:56:35 -07:00
Co Tran
79b1c5415f
[Kan-103] add support toxic/jailbreak model ( #49 )
...
* add toxic/jailbreak model
* fix path loading model
* fix syntax
* fix bug,lint, format
* fix bug
* formatting
* add parallel + chunking
* fix bug
* working version
* fix onnnx name erorr
* device
* fix jailbreak config
* fix syntax error
* format
* add requirement + cli download for dockerfile
* add task
* add skeleton change for envoy filter for prompt guard
* fix hardware config
* fix bug
* add config changes
* add gitignore
* merge main
* integrate arch-guard with filter
* add hardware config
* nothing
* add hardware config feature
* fix requirement
* fix chat ui
* fix onnx
* fix lint
* remove non intel cpu
* remove onnx
* working version
* modify docker
* fix guard time
* add nvidia support
* remove nvidia
* add gpu
* add gpu
* add gpu support
* add gpu support for compose
* add gpu support for compose
* add gpu support for compose
* add gpu support for compose
* add gpu support for compose
* fix docker file
* fix int test
* correct gpu docker
* upgrad python 10
* fix logits to be gpu compatible
* default to cpu dockerfile
* resolve comments
* fix lint + unused parameters
* fix
* remove eetq install for cpu
* remove deploy gpu
---------
Co-authored-by: Adil Hafeez <adil@katanemo.com>
2024-09-23 12:07:31 -07:00
