Michael Günther @michael_g_u

ML @jinaai_ github.com/guenthermi Berlin, Germany Joined August 2022

Tweets

290
Followers

621
Following

214
Likes

271

Jina AI @JinaAI_

3 weeks ago

V4 is multimodal embeddings, but V4-GGUF wasn't—until now. We've finally cracked how to generate multimodal embeddings using llama.cpp & GGUF. We fixed two main issues. First, in the language model part, we corrected the attention mask in the transformer block so it properly…

7 33 171 12K 91

Download Image

𝚐𝔪𝟾𝚡𝚡𝟾 @gm8xx8

3 weeks ago

mmBERT: Massively Multilingual BERT Trained on 3T+ tokens across 1,833 languages, mmBERT surpasses XLM-R on standard NLU and retrieval benchmarks and is competitive with English-only encoders; in throughput tests it runs 2–4× faster than prior multilingual encoders under…

𝚐𝔪𝟾𝚡𝚡𝟾 @gm8xx8

2 months ago

1 1 8 6K 5

Download Image

3 10 55 5K 23

Download Image

Jina AI @JinaAI_

4 weeks ago

Today we're releasing jina-code-embeddings, a new suite of code embedding models in two sizes—0.5B and 1.5B parameters—along with 1~4bit GGUF quantizations for both. Built on latest code generation LLMs, these models achieve SOTA retrieval performance despite their compact size.…

9 51 314 29K 206

Download Image

Michael Günther @michael_g_u

a month ago

We are at @qdrant_engine 's Vector Space Day 🚀 in Berlin on Sep 26. We'll talk about "Vision-Language Models: A New Architecture for Multi-Modal Embedding Models" and also share some insights and learnings we gained while training jina-embeddings-v4. 🎫 lu.ma/p7w9uqtz

0 2 6 400 2

Download Image

Jina AI @JinaAI_

a month ago

Got a Mac with an M-chip? You can now train Gemma3 270m locally as a multilingual embedding or reranker model using our mlx-retrieval project. It lets you train Gemma3 270m locally at 4000 tokens/s on M3 Ultra - that's actually usable speed. We've implemented some standard…

7 65 424 31K 412

Download Image

Jina AI @JinaAI_

2 months ago

Two weeks ago, we released jina-embeddings-v4-GGUF with dynamic quantizations. During our experiments, we found interesting things while converting and running GGUF embeddings. Since most of the llama.cpp community focuses on LLMs, we thought it'd be valuable to share this from…

5 25 174 12K 122

Download Image

Michael Günther @michael_g_u

2 months ago

I went together with @bo_wangbo to SIGIR this year, we wrote a blog post with our highlights and summaries of AI and neural papers that we found interesting at the conference jina.ai/news/what-we-l…

0 1 10 2K 6

Jina AI @JinaAI_

2 months ago

Our official MCP server with read, search, embed, rerank tools on mcp[at]jina[at]ai, where we optimized the embedding and reranker usage particularly for context engineering for LLMs.

7 17 140 26K 91

Download Image

tomaarsen @tomaarsen

2 months ago

😎 I just published Sentence Transformers v5.1.0, and it's a big one. 2x-3x speedups of SparseEncoder models via ONNX and/or OpenVINO backends, easier distillation data preparation with hard negatives mining, and more! See 🧵for the deets:

1 15 132 5K 43

Download Image

Michael Günther @michael_g_u

2 months ago

Resolution is important for image embeddings - especially for visual document retrieval. jina-embeddings-v4 supports inputs up to 16+ MP (the default is much lower). We wrote a blog post about how resolution affects performance across benchmarks jina.ai/news/how-image…

0 2 11 427 2

Bo @bo_wangbo

2 months ago

Finally, a 45 page literature review of text embedding model, datasets, evaluation and training methods: arxiv.org/abs/2507.20783

0 52 364 19K 325

Michael Günther @michael_g_u

2 months ago

We created a new benchmark for visual document retrieval with diverse visually rich documents (more than linear paginated PDFs) and more query types than just questions github.com/jina-ai/jina-v…

Jina AI @JinaAI_

2 months ago

We created a new benchmark for visual document retrieval with diverse visually rich documents (more than linear paginated PDFs) and more query types than just questions github.com/jina-ai/jina-v…

4 19 121 11K 66

Download Image

0 0 6 524 1

Felix @felix1987_

2 months ago

vLLM is finally supporting our multi-modal reranker jina-reranker-m0 huggingface.co/jinaai/jina-re… This is neat!

0 2 12 798 1

Download Image

elie @eliebakouch

2 months ago

We've just release 100+ intermediate checkpoints and our training logs from SmolLM3-3B training. We hope this can be useful to the researcher working on mech interpret, training dynamics, RL and other topics :) Training logs: -> Usual training loss (the gap in the loss are due…

13 60 401 32K 189

Download Image

Jina AI @JinaAI_

2 months ago

jina-embeddings-v4-GGUF is here with different quantizations github.com/jina-ai/jina-e… Unsloth-like dynamic quants is on the way.

1 23 133 8K 58

Jina AI @JinaAI_

3 months ago

Context engineering is curating the most relevant information to pack the context windows just right. Text selection and passage reranking are integral components of it. In part 2 of our Submodularity Series, we show that both text selection and passage reranking yield to…

6 16 146 11K 134

Download Image

Michael Günther @michael_g_u

3 months ago

We just arrived @SIGIRConf! If you're here or are interested in an internship @JinaAI_ on training the following search foundation models, feel free to reach out to me: - Embedding / Dense Retrieval Models - Rerankers - Small LMs (<2B) for document cleaning, extraction, etc.

0 4 33 2K 4

Download Image

Michael Günther @michael_g_u

3 months ago

Our paper "Late Chunking: Contextual Chunk Embeddings Using Long-Context Embedding Models" has been accepted at the Robust IR Workshop @ SIGIR 2025! 🌠 📅 I'll present it on July 17th 📝 Pre-print: arxiv.org/abs/2409.04701 🔗 Workshop: …-2025-workshop-on-robust-ir.github.io