Leon Engländer

14

4.7K

Leon Engländer retweetledi

Clifton Poth@clifapt·2 Mar

Took Claude up for a spin on the weekend and started a quick open-source self-hosted re-implementation Thinking Machines' Tinker API: github.com/calpt/open-tin…

English

3

7

304

Leon Engländer@LeonEnglaender·21 May

@derguene @PfeiffJo We also have composition blocks for combining adapters (like stacking or multi-task learning) besides plain adapter merging. Plus, methods like bottleneck adapters & PyReFT (and their variants), which go beyond what's standard in PEFT, giving you more building blocks for research

English

2

23

Leon Engländer@LeonEnglaender·21 May

@derguene @PfeiffJo Hi @derguene, as one of the maintainers of Adapters, I can answer this🙂Adapters is more specialised towards research: We focus on making it easy to modify/create methods & offer more flexibility. For instance, you can add multiple heads to a model & switch them out (1/2)

English

1

0

2

26

Jonas Pfeiffer@PfeiffJo·21 May

Check it out! The newest version of AdapterHub is on 🔥

AdapterHub@AdapterHub

🚀Adapters v1.2 is out!🚀 We've made Adapters incredibly flexible: Add adapter support to ANY Transformer architecture with minimal code! We used this to add 8 new models out-of-the-box, incl. ModernBERT, Gemma3 & Qwen3! Explore this +2 new adapter methods in this thread👇(1/5)

English

16

1.2K

Leon Engländer@LeonEnglaender·21 May

Thrilled about our new Adapters release!🎉I had a blast working on this version, especially contributing to the new plugin interface (like adding ModernBERT) and helping with the VeRA adapter method. Have a look at the full thread for all the awesome updates from our team 👇

AdapterHub@AdapterHub

🚀Adapters v1.2 is out!🚀 We've made Adapters incredibly flexible: Add adapter support to ANY Transformer architecture with minimal code! We used this to add 8 new models out-of-the-box, incl. ModernBERT, Gemma3 & Qwen3! Explore this +2 new adapter methods in this thread👇(1/5)

English

1

3

546

Leon Engländer retweetledi

AdapterHub@AdapterHub·30 Oca

🎁 A new update of the Adapters library is out! Check out all the novelties, changes & fixes here: github.com/adapter-hub/ad…

English

4

5

626

Leon Engländer retweetledi

UKP Lab@UKPLab·7 Kas

🎉M2QA has been accepted to #EMNLP Findings!🎉 M2QA is a new multilingual and multidomain QA dataset. We show that current transfer methods are insufficient and that language & domain transfer aren't independent! 📄 Paper: arxiv.org/abs/2407.01091 👇👇👇 twitter.com/LeonEnglaender…

📢 New preprint 🎉 We introduce "M2QA: Multi-domain Multilingual Question Answering", a benchmark for evaluating joint language and domain transfer. We present 5 key findings - one of them: Current transfer methods are insufficient, even for LLMs! 📜arxiv.org/abs/2407.01091 🧵👇

English

2

15

843

Leon Engländer retweetledi

Hannah@h_sterz·27 Eyl

Do you DARE? Introducing a multiple-choice VQA benchmark ✨DARE✨ with: - 4 main robustness evaluation ⛓️ - 5 diverse categories 🧩 - Extensive analysis of 4 widely used VLMS 🤖

English

1

7

15

2.7K

Leon Engländer retweetledi

UKP Lab@UKPLab·21 Eyl

@emnlpmeeting @IGurevych @TUDarmstadt @ProLOEWE @ATHENECenter @emergen_CITY @CS_TUDarmstadt @FraunhoferSIT @Hessian_AI @jtonglet @Furkansahinuc @thy2512 @yufanghou @IBMResearch @AmazonScience @Koby_Loby @XiaoL558286 @ndaheim_ @dmacjam @mrinmayasachan @trumancfy @xinranz3 @TongChen0 @soshsihao @hongming110 @koeppl_lab @CarnegieMellon @uwcse @AndreasWaldis @anne_lauscher @DS_research_HH @hslu @summetix @HaritzPuerto @mtutek @somakaditya @XiaodanZhu2048 @IITKgp @IngenuityLabs @QueensECE @QianRuan_ @ilokuznetsov One more paper is accepted to the Findings of #EMNLP2024: »M2QA: Multi-domain Multilingual Question Answering« by @LeonEnglaender, @h_sterz, @clifapt, @PfeiffJo, @ilokuznetsov, @IGurevych 📰arxiv.org/abs/2407.01091 💻github.com/UKPLab/m2qa (11/🧵)#EMNLP2024

English

5

9

1K

Leon Engländer@LeonEnglaender·12 Ağu

We've just released v1.0 of Adapters!🎉 A lot has happened since the original release of Adapters. Major highlight: Adapter Merging, which opens up fascinating possibilities for transfer learning – hence it's at the heart of my current research. Stay tuned for upcoming papers!🔬

AdapterHub@AdapterHub

🎉Adapters 1.0 is here!🚀 Our open-source library for modular and parameter-efficient fine-tuning got a major upgrade! v1.0 is packed with new features (ReFT, Adapter Merging, QLoRA, ...), new models & improvements! Blog: adapterhub.ml/blog/2024/08/a… Highlights in the thread! 🧵👇

English

11

262

Leon Engländer retweetledi

AdapterHub@AdapterHub·12 Ağu

🎉Adapters 1.0 is here!🚀 Our open-source library for modular and parameter-efficient fine-tuning got a major upgrade! v1.0 is packed with new features (ReFT, Adapter Merging, QLoRA, ...), new models & improvements! Blog: adapterhub.ml/blog/2024/08/a… Highlights in the thread! 🧵👇

English

7

45

5.4K

Leon Engländer retweetledi

Clifton Poth@clifapt·2 Tem

Check out Leon's first first-author paper! M2QA for cross-domain & cross-lingual transfer in QA ⬇️

📢 New preprint 🎉 We introduce "M2QA: Multi-domain Multilingual Question Answering", a benchmark for evaluating joint language and domain transfer. We present 5 key findings - one of them: Current transfer methods are insufficient, even for LLMs! 📜arxiv.org/abs/2407.01091 🧵👇

English

1

4

464

Leon Engländer retweetledi

Hannah@h_sterz·2 Tem

Checkout our new Multi-Domain Multilingual Question Answering Benchmark ✨M2QA✨ for joint language and domain transfer 🚀 ⬇️⬇️⬇️

📢 New preprint 🎉 We introduce "M2QA: Multi-domain Multilingual Question Answering", a benchmark for evaluating joint language and domain transfer. We present 5 key findings - one of them: Current transfer methods are insufficient, even for LLMs! 📜arxiv.org/abs/2407.01091 🧵👇

English

1

7

513

Leon Engländer retweetledi

Jonas Pfeiffer@PfeiffJo·2 Tem

Check out our new multilingual and multidomain QA dataset 👇

📢 New preprint 🎉 We introduce "M2QA: Multi-domain Multilingual Question Answering", a benchmark for evaluating joint language and domain transfer. We present 5 key findings - one of them: Current transfer methods are insufficient, even for LLMs! 📜arxiv.org/abs/2407.01091 🧵👇

English

1

21

1.2K

Leon Engländer retweetledi

AdapterHub@AdapterHub·2 Tem

📢 New preprint 🎉 We - the AdapterHub team - present the M2QA benchmark to evaluate joint domain and language transfer! 🔬 Key highlight: We show that adapter-based methods on small language models can reach the performance of Llama 3 on M2QA! 🚀 👇

📢 New preprint 🎉 We introduce "M2QA: Multi-domain Multilingual Question Answering", a benchmark for evaluating joint language and domain transfer. We present 5 key findings - one of them: Current transfer methods are insufficient, even for LLMs! 📜arxiv.org/abs/2407.01091 🧵👇

English

2

8

616

Leon Engländer@LeonEnglaender·2 Tem

Work done with @h_sterz, @clifapt, @ilokuznetsov, @PfeiffJo and @IGurevych 📜 Paper: arxiv.org/abs/2407.01091 🌐 Code: github.com/UKPLab/m2qa We uploaded the M2QA dataset and adapters to @huggingface 🤗 M2QA dataset: huggingface.co/datasets/UKPLa… 🤗 Adapters: huggingface.co/collections/Ad…

English

3

95

Leon Engländer@LeonEnglaender·2 Tem

Overall, we see that Aya23 and gpt-3.5 perform best and - equally exciting - we see that our adapter-based methods can surpass the performance of Llama 2 and reach the performance of Llama 3 even though they only build on the much smaller model XLM-RoBERTa!

English

1

0

1

89

Leon Engländer@LeonEnglaender·2 Tem

📢 New preprint 🎉 We introduce "M2QA: Multi-domain Multilingual Question Answering", a benchmark for evaluating joint language and domain transfer. We present 5 key findings - one of them: Current transfer methods are insufficient, even for LLMs! 📜arxiv.org/abs/2407.01091 🧵👇

English