Skip to content

Adds local LLM for mistral nemo via Ollama #170

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 102 commits into
base: main
Choose a base branch
from
Open

Adds local LLM for mistral nemo via Ollama #170

wants to merge 102 commits into from

Conversation

@DanielMarchand
Copy link

No description provided.

drudilorenzo and others added 30 commits April 26, 2024 15:04
@drudilorenzo
fix(backend): replace deprecated models.
3da4834
@drudilorenzo
feat: simplify execution w/ scripts.
c7d7968
@drudilorenzo
feat: add cost logging.
0de1b36
@drudilorenzo
fix: fix scripts.
09d7ae2
@drudilorenzo
feat: cost logging improvement.
bb269be
@drudilorenzo
feat: add cost logger singleton.
c7932d9
@drudilorenzo
feat: thread-safe singleton.
afb52d8
@drudilorenzo
fix: remove clustrmaps.
ece4f11
@drudilorenzo
fix: create movement directory.
11e4aaa
@drudilorenzo
feat: improve project config.
67ff092
@drudilorenzo
tbd: add current bug.
7a599b5
@drudilorenzo
feat: untrack temp file.
735c15e
@drudilorenzo
Add readme.
9c8d2f6
@drudilorenzo
Update README.md
3e93bde
@drudilorenzo
Update README.md
433fdd2
@drudilorenzo
Update README.md
11f385e
@drudilorenzo
(docs): improve readme.
4dda6d2
@drudilorenzo
feat: log on file.
4a36e0b
@drudilorenzo
feat: create directory.
b71edf6
@drudilorenzo
(exec): add skip-morning simulation.
e017db1
@drudilorenzo
(scripts): improve backend scripts.
b4f59aa
@drudilorenzo
feat: restart execution in case of failure.
e8a73ad
@drudilorenzo
(docs): improve readme.
b9f160f
@drudilorenzo
Update README.md
f1cec26
@drudilorenzo
Update README.md
5b61652
@Laz4rz
add tracing func, add fake get for frontend refresh, add data files
012fe0e
@Laz4rz
add setup info
69e5a6c
@drudilorenzo
feat: save checkpoints.
39aa57b
@Laz4rz
add test elections setup with 5 voters
8ae36ef
@Laz4rz
fix base files in .gitignore
fae94f0
@DanielMarchand
udpated hourly schedule prompt for better results
711e2e6
@DanielMarchand
added note about the embeddings not working on n25
cd6c080
@DanielMarchand
prompt changes for llama3 and cleaning up the task decomp function
c307729
@DanielMarchand
updates to sections that generate (s,p,o) triples)
f87c65d
@DanielMarchand
substantial improvements in local run time
d2201b9
@DanielMarchand
local model very close to being able to run independantly for lengthy…
361607c 
… period of time
@DanielMarchand
made task decomp regex more robust, set min temperature to 0.1 in all…
5f76763 
… tasks
@DanielMarchand
fixing issues with generating focal point
ed07874
@DanielMarchand
refactoring event triple, object event and insight and evidence, code…
d276734 
… is becoming relatively stable
@DanielMarchand
fixed instabilities in room locator, gave ChatGPT_safe more retries
bb20209
@DanielMarchand
removed misleading comments (probably copy/pasted from another function)
559ceea
@DanielMarchand
substantial work to generating triples, as well as location and decom…
8b5e12f 
…p schedule prompts
@DanielMarchand
restored to only using the debug variable in utils
983b34b
@DanielMarchand
added EXCEPT_ON_FAILSAFE=False default to utils.py (preserves default…
a1dbd26 
… behaviour)
@DanielMarchand
removed symbol in emojii (may cause problems on windows though?)
5a61885
@DanielMarchand
removed some dead comments and print statements
0fede96
@DanielMarchand
except on too many retries set to False
5181a3a
@DanielMarchand
added simulation name validator so nobody else has to waste an hour t…
4d6e5d1 
…rying to figure out Django view exceptions
@DanielMarchand
Merge commit '4d6e5d18' into fix-and-improve_add-ollama
644005f
@DanielMarchand
fixed issue in merge
8dee870
@DanielMarchand
improved parsing of some outputs, code is now largely stable for loca…
c1ed7fc 
…l llm
@DanielMarchand
made compress sim use a command line argument
b08ea5c
@DanielMarchand
made the safe_generate command easier to parse the complete trace int…
1da280d 
…o a json for later model checking and evaluation
@DanielMarchand
switching to mistralnemo llm
10bc2c0
@DanielMarchand
finally replaced all ChatGPT_safe_response with safe_response, fine-g…
561c9c7 
…raining the use of triples
@DanielMarchand
fixed bug in logging that would mess up the retry count
8907061
@DanielMarchand
poignancy more robust to errors, failsafe set to 1
c5d6351
@DanielMarchand
moved nemo prompts to the other repos, due to parser changes this is …
9d49a96 
…the only one I know that 'should' work
@DanielMarchand
place the extracted json files into separate directories
004bf75
@DanielMarchand
set max number of restarts to be 10 else risk of filling disk
dd05d70
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@DanielMarchand @drudilorenzo @Laz4rz @ketsapiwiq