Daniel

WIP Spec

bug: Unable to detect manually imported Model

5

**Describe the bug** **To Reproduce** Steps to reproduce the behavior: - JasonB from Discord is unable to get Jan to detect manually imported model (from creating folder and pasting GGUF...

**Describe the bug** **To Reproduce** Steps to reproduce the behavior: 1. Go to Hub 2. Download Phi-2-3b or Mixtral 3. Sometimes, the CDN for Huggingface will error out in the...

epic: Jan has Conversation-based RAG

6

# Objectives - Allow users to upload text files (PDFs, docs, etc.) within the chat interface. - Deliver a user-friendly file upload mechanism, enabling seamless integration into the chat environment....

feat: Models should download to a user-defined Directory

2

**Problem** - We currently download models to an app directory (e.g. `Application Support` for Mac) - Problems - User cannot add/delete or copy/paste their own models in - Very dependent...

feat: Threads can override and inherit default params from Assistants, Models, and Engine

11

## Tasks - [x] Eng Spec @dan-jan - [ ] Local Models - [ ] Remote Models (KIV) - [x] janhq/internal#66 - Functionality - [x] Each Recommended Model should have...

feat: UI Framework for Extensions

1

## Objective - Extensions can register UI elements - Our UI is mainly driven by Extensions (vs. hard-coded) ## Inspirations - https://docs.obsidian.md/Home - https://code.visualstudio.com/api/ux-guidelines/overview ## Todos - [x] #776

epic: Jan has better Hardware Monitoring (e.g. Prometheus Exporter, etc)

2

## Kanban https://github.com/orgs/janhq/projects/5/views/7?filterQuery=milestone%3A%22Jan+has+System+Monitor%22 ## System Monitor _Questions_ - [ ] Are we able to do a Prometheus/Grafana-style time series dashboard? (How do we even persist the data on flat files?)...

feat: Thread titles should auto-summarize Topic

7

 Our current Jan results in "New Thread" for each thread, as users don't title each thread manually. We should find a way to implement a ChatGPT-style Thread Title summarizer,...

feat: enable `continue` button for when LLM responses exceed token limit

11

**Describe the bug** LLM responses seem to terminate prematurely, do we have a max reply token setting somewhere? EDIT: Mistral likely limited to 512? **To Reproduce** Steps to reproduce the...