Fix JSON response parsing in ux_testing_fsm function to directly access the response object.

added ux_testing api

Enhance answer generation by adding optional context parameter to relevant functions and update .gitignore to include test directory.

Clarify verbosity guidelines in prompts to avoid unnecessary elaboration.

using litellm - for now shuffle disabled

Update requirements to include TinyTroupeFork repository

fix

trying with english

forced language of questions in prompts

forced focus on questions in prompts

temperature to 0.5 for answers

now using 4.1mini with temperature 1

fixed models and temp for prod

removed parameters from eval script

fix on persona generation

removed tinytroup from requirements for simplicity

restored function for parameters generation - just for completeness

removed parameters from generation personas api

added possibility to ask questions in bulk - now default

modified prompts for ask question and build personas - fixed temperature to 0 and model to 4.1 nano for testing

modified persona generation logic: now fixed set of parameters

fix on answers

added function to save also users

prompts in a separated file

explicit parameters in call_llm function

added variability in synth answers

modify n of fleet in test script

fixed n of parameters

modified gitignore

complete eval flow ready

using direct embedding comparison

added pairwise approach

added eval script and data

fix in utils

fix on split

splitted users batch generation and answer generation

added generate-users-answers endpoint

increased to 20 params per user

added language in prompt for chat with persona

fix prompt report

all questions and users go in parallel when asking questions

fixed call_llm and models in apis

added test script

cleaned code - improved response format - splitted generate users adding new api for parameters

trying gpt nano 4.1

refined prompts in dt flow

increased max personas to generate

now returning only interactions of the assistant in chat-with-twin

fix save person api

modified reset agents