Improve the prompts

[StephanAkkerman]

Description

This PR [adds/removes/fixes/replaces] the [feature/bug/etc].

Related Tickets & Documents

Fixes # (issue)

What type of PR is this? (check all applicable)

• 🍕 Feature
• 🐛 Bug Fix
• 📝 Documentation Update
• 🎨 Style
• 🧑‍💻 Code Refactor
• 🔥 Performance Improvements
• ✅ Test
• 🤖 Build
• 🔁 CI
• 📦 Chore (Release)
• ⏩ Revert

Added to documentation?

• 📜 README.md
• 📓 Wiki
• 🙅 No documentation needed

[code-review-doctor]

Some food for thought. View full project report here.

[Copilot]

Pull Request Overview

This PR refactors prompt generation and model configuration, adds new utilities for syllabification and n-gram probabilities, and updates dependencies and pipelines.

• Updated config.yaml with new LLM/IMAGE_GEN model names and parameters, including FLUX LoRA settings
• Expanded requirements.txt with language‐processing libraries and platform‐specific Nunchaku wheels
• Introduced syllabifier.py and ngram_probs.py; refactored VerbalCue and ImageGen services for improved prompt templates and FLUX integration

Reviewed Changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
config.yaml	Swapped model names; added MNEMONIC_CANDIDATES, params
backend/requirements.txt	Downgraded transformers; added pyphen/lingpy/epitran, nunchaku wheels
backend/mnemorai/utils/syllabifier.py	New language syllabification utility
backend/mnemorai/utils/ngram_probs.py	New n-gram probability utility with corpus downloads
backend/mnemorai/services/pre/translator.py	Improved logging around language code mapping
backend/mnemorai/services/imagine/verbal_cue_gen.py	Refactored mnemonic/verbal-cue prompt templates
backend/mnemorai/services/imagine/image_gen.py	Replaced SanaPipeline with FLUXPipeline; seed & offload logic
backend/mnemorai/run.py	Updated example invocation

Comments suppressed due to low confidence (3)

config.yaml:33

• Ensure that MNEMONIC_CANDIDATES is correctly indented under the LLM section and consider documenting this new parameter in the README so users know how to adjust it.

MNEMONIC_CANDIDATES: 20

backend/mnemorai/services/imagine/image_gen.py:57

• The old SanaPipeline branch was removed; ensure that any users or docs referencing Sana pipelines are updated and that the default branch (AutoPipelineForText2Image) still covers all intended models.

if "flux" in self.model_name.lower():

backend/requirements.txt:4

• Downgrading Transformers from 4.52.4 to 4.51.3 may reintroduce bugs or security issues; please confirm that this change is intentional and update any related compatibility tests or documentation.

transformers[sentencepiece]==4.51.3