v0.10.0

0.10.0 (2024-11-09)

Features

• article: add section about redlm deep dive video (5ed0c85)
• article: expand section about llamaindex and add paragraph on LLMRerank (7dac701)
• article: reword article, add links (03384f0)
• docker: update docker and docker compose configuration and documentation in readme (5c6c80e)
• nginx: add nginx config for running api and ui on same domain (9af720c)
• rag: rector CustomQueryEngine, add observability with langfuse, fix issues with PromptTemplates and RAG logic (0133cd3)
• readme: update readme instructions and add section about redlm features (0a3d072)
• workflow: fully integrate llama-indeex workflows into FastAPI application (5003174)
• workflows: add workflow for rag Q&A bot using rerank (bccd6e6)

Bug Fixes

• black: format code with black (a7c7b35)
• typo: fix typo in logging (1e5427b)
• workflows: filter index by chapter when processing image Q&A requests (9182f3b)

v0.9.0

0.9.0 (2024-11-04)

Features

• article: add section about models used in RedLM (bff2f70)
• article: draft of final thoughts (fa9d13f)
• article: update article with more examples (b92215e)
• article: update article with more examples and images (ef96491)
• index: use both english and chinese embed models during indexing (25c4135)
• index: use both english and chinese embed models during indexing (cab16ec)
• language: support for Q&A queries in English and Chinese with dynamic prompts, update article, add examples (780b5a7)
• milvus: add support for using external milvus server with docker compose via env var configuration (872a2b0)

Bug Fixes

• typo: fix all typos in article using spellcheck extension cSpell (3f151c0)
• ui: fix issue with cn function from shadcn and radix (ce11c5d)

v0.8.0

0.8.0 (2024-10-22)

Features

• article: add article complete rough draft in markdown format with images (c3741df)
• article: rtx pc cluster, tailscale, notebooklm, cloudflare tunnels (c4af786)
• k8s: add qwen2-vl service kubernetes config using kustomize (738afd3)
• kustomize: add llm service to kustomize configuration (ff84540)

v0.7.0

0.7.0 (2024-10-05)

Features

• article: add initial draft of article and images (3413882)
• docker: add docker config config for ui app (0585afd)
• docker: add docker config for redlm-api fastapi backend (cc0ac00)
• docs: update readme with instructions on how to start inferences services locally for llm and vlm (cab78c6)

Bug Fixes

• llm: add max_tokens to local llm config and LLM_NAME env var (063d2ae)

v0.6.0

0.6.0 (2024-10-01)

Features

• data: add json files with chapter data, text and translations (854a619)
• name: rename hllm directory to redlm (2d3f662)

Bug Fixes

• git: add .gitkeep to ui/public/img to keep folder (dd9e1b8)
• git: add .gitkeep to ui/public/img to keep folder (74367b1)

v0.5.0

0.5.0 (2024-09-30)

Features

• evaluation: update logging in scripts for evaluation (e6e9606)
• multi-modal-rag: finish implementation of multi-modal rag (c38a356)
• nims: use NIMs for LLM inference (53b49ab)
• nvidia: finish implementation of nvidia cloud apis for image q&a with vision language models (f8ff0b3)
• translation: improve logging for translation script (7aa55b4)
• translation: translated all 120 chapters successfully, fixed context window errors, tuned LLM parameters (fb05ed5)
• translation: update script for translation (aad58dd)
• translation: update translation script and translation prompts, add new translation using Qwen2-7B base model (6978bd4)
• ui: better styles for reference badges and image selection (8934f5e)

Bug Fixes

• lint: format code with black (12b1835)
• ui: formatting for mmqa ui (49d012a)

v0.4.0

0.4.0 (2024-09-12)

Features

• metadata: return node metadata from query engine and display in ui (ff7eab3)
• multi-gpu: changes to support distributed LLM generation (296e774)
• multi-modal-ui: add formatting to image chat form and text boxes (b22390b)
• multi-modal: add endpoint and test script for multi-modal Q and A (87baeb7)
• multi-modal: add pinia store for multi-modal data and refactor image page to use script setup syntax (96e4dea)
• q-and-a-ui: improve formatting for q and a component, fix navigation errors (a0807d9)
• rag: add additional endpoint for question and answer (63aab04)
• translation: add mandarin and english translations to chapter detail page (00b9b0a)
• translation: add translations made with qwen2-7b-chat (5051deb)
• translations: add additional translations using qwen2-7b-chat (d29b51a)
• trt-llm-api: use tensorrt-llm api to do translation, add sample translation for chapter 111 (3cca5b6)
• ui: add modified drawer component for displaying chat message responses (befa2b0)
• ui: fix fastapi and ui integration issues, add Q and A page (51d4acb)

Bug Fixes

• lint: format with black (991664b)
• ui: scroll to bottom after question is answered (833a45a)

v0.3.0

0.3.0 (2024-09-08)

Features

• black: format code with black (39c1342)
• index: refactor RAG server, add command to build and persist VectorIndexStore data (62405f9)

Bug Fixes

• black: format with black (90725e1)

v0.2.0

0.2.0 (2024-09-07)

Features

• img: rename images (dbb7cbe)
• ui: add pinia for chapter store, add components on index page (c4975ed)
• ui: updates to UI for reading text and viewing images (10ddabe)

v0.1.0

0.1.0 (2024-09-07)

Features

• baichuan: add service for baichuan2 (be077f4)
• eval: improve evalution scripts and optimize prompts for better eval score (f754c88)
• img: add scaled down images of paintings (0066b55)
• python: python env (3e43f4f)
• qwen2-vl: add qwen2-vl service for multimodal LLM (769d387)
• qwen2-vl: fixes for qwen2-vl service, example client API call now working (8f5af85)
• rag: add basic rag example in a sample script (c9c7efa)
• rag: add fastapi server for serving llama index rag app (b2e72ac)
• release: add release please for changelog (7713e90)
• scripts: add scripts for text processing, evaluation, components for displaying text and cropping images (df273ba)
• text: add text and scraping script (eb8dc55)
• ui: add basic ui with nuxt (93da4c1)
• vue: add shadcn helper functions and component code (dfe04db)