Commit History
-
adjusting prompts
d3ae060 -
Updating the instruction's order
updating system prompt.5f7776b
-
rolling back
1b550e9 -
updating prompts for recommendations
updating follow up prompts updating drumeo banned words and phrasesb505cb8 -
updating prompts for recommendations
updating follow up prompts updating drumeo banned words and phrases007bbac
-
adding more banned contents
modifying prompts91a7fde -
adding more banned contents
0ee6a35 -
Revert "- modifying system prompts to have CTA"
This reverts commit e1ed1b8cbdb37bf2299a8b1a2a8216d65709b7e4.2ff618f -
- modifying system prompts to have CTA
- modifying prompts for recommended contents for better references. - adding new items to banned listse1ed1b8
-
- updating banned list of contents
- changing selection logic, following the logic that we have in productione6f81c2 -
- updating banned list for guitareo
c309583 -
- Adding banned list for guitareo
- adjusting prompts35a1eaf
-
- Adding csv files for each brand to speed up the loading time
65016d2
-
- updating the system to be able to work with all brands
ce32f1f
-
- reading users directly from campaign (proxy based on snowflake logics)
707d2d4 -
- Adding time reference instructions
080e8a0 -
- adding price analysis per model
6f416a9 -
- adding robust parsing strategy for LLM output
- including token counter for all models for price analysis18fc113 -
- Fix: fetching user info via their email as identifier
aa19464 -
- error handling: be able to select other brands if the config file doesn't exist yet.
d017932
-
- fixing bug for sequential messages
74c9bd0 -
- fixing bug
9ade43d -
- Handling error more efficiently
- removing first name - adjusting preferences9b1726d
-
- increasing max_try for request
7ba4670 -
- adding other LLMs
d59e3c0
-
- modifying cost for request
- keeping reliable LLMs282b49d -
- modifying instructions
7d1fbef -
- Adding recommendation instruction
b864e48 -
- Adding recommendation for each step
- Random recommendation from top results - fixing bug and making the system more robust - modifying prompts77c4795
-
modifying prompts
1c00bf4 -
modifying prompts
improving the reliabilityb942b6b
-
Changing system prompts and modifying the instructions.
b243899 -
Changing system prompts and modifying the instructions.
37fd8e1 -
Changing system prompts and modifying the instructions.
4d858d6
-
fixing bug
b53741d
-
creating the first version
f474755 -
Creating the project
- changing UI to get examples and segment info - fetching user profiles to choose the best options - adding recommendations - adjusting instructions and prompts53d317f
-
- adjusitng prompts
- fixing bugs and keeping only better LLMs to keep the UI cleaner8ad6512
-
- ignoring user_profiles and first names
- adjusitng prompts6fe328f
-
- adding user_profile for test
- connecting to recsys_v3 results - fixing role to read data properly3cb83db -
- Fixing MultiMessage.py
2862dc5 -
-changing subsequent messages similar to initial messages and change prompting
- using user profiles for personalizationfa44311
-
-by passing protection layer and prompt Eng layer to simplify the process.
- ignoring negative examples and banned words4f26c60
-
Adding positive reinforcement in system prompt and modifying prompts
b783618
-
Fixing bugs
bea601a -
Fixing bugs
a4d8b4b -
Fixing version conflicts
6b7de2a -
Adding Google Gemini 2.5 flash lite
52ee99e
-
- deleting Qwen due to unexpected behavior
2191b8a -
- Adding inference APIs
Adding below models: Gemma 3 - 27B Llama 3.2 - 11B Qwen 2.5 - 7B04f2d25