Top Tweets for #Longcontext
๐ Multimodal AI: Process text, images, audio, and video
๐ง Long-context: Handle complex, large inputs
๐ค Agentic AI: Plan tasks with less human input
#AITrends #MultimodalAI #LongContext #AgenticAI
https://t.co/G7y0woyS8Q
the wild part is that the 20B model got a +4.4% bump just by training on its own generated reasoning traces โ basically selfโteaching. All from Wikipedia tables and a dualโpath code verifier. ๐คฏ #LLM #longcontext
https://t.co/X6AeJ2AHPZ
BEAVER ร OpenClaw ๐ฆซ๐ฆ
/beaver What is the main contribution?
From now on, long texts will no longer burn tokens ๐ธ
One command: long text compression โ question โ answer, fully automated.
https://t.co/gY2tVGlCo9
@steipete
#Longcontext #LLM #openclaw
๐ FlagOS Open Computing Global Challenge โ Track 3
Automatic data annotation for LLMs in ultra-long contexts, using long-context ICL and multi-step reasoning with FlagScale
๐ฐ ยฅ140,000 prize pool | Global
https://t.co/HOCSDtW4Q4
#FlagOS #LLM #LongContext #Competition
๐GLMโ4.7 is live on CanopyWave Cloud! A bilingual frontier model for English and Chinese, upgraded with stronger reasoning and context handling.
๐ Try it now:https://t.co/pk9NuYXm8h
ยท More info: https://t.co/lAFCEqCJOh
#AI #Bilingual #Reasoning #LongContext #GLM #MODEL #TOKEN
5/5
The tools for the next leap in AI are being forged right now.
This is one to watch closely.
Get on the list and see where it goes.
๐ https://t.co/71QDQ5cLRV
#AI #LongContext #GenerativeAI #BUIDL #AGI
Today, Anthropic revealed that their Claude models can now handle way longer prompts.
think Lord of the Rings trilogy or 75,000 lines of code in a single go. Mind. Blown. ๐ฒโจ
#AI #MachineLearning #TechNews #Anthropic #ClaudeAI #FutureOfAI #LongContext #NextGenAI
๐ Discover how long-context LLMs and RAG work together to build scalable, efficient and intelligent AI systems for enterprise use.
๐ Read the full blog: https://t.co/6noN9mL3zM
#Floatbot #LLMs #RAG #LongContext #AI #News #GenAI #AgenticAI #AIResearch
#LongContext / document-level MT #QualityEstimation model, based on #ModernBERT. It is trained on a dataset of long-context translation pairs with up to 32 sentences (64 sentences in the source and target).
#NLProc #MachineTranslation
https://t.co/xpUNBL28az
There is a lack of a clear definition of #LongContext, as studies vary in their token thresholds (e.g., 8k, 32k, 128k). LC datasets can also be realistic (e.g., novels, research papers) and synthetic (e.g., concatenated Wikipedia excerpts) long texts. Realistic texts align with comprehension tasks, while synthetic texts focus on factual reasoning. The challenges of defining #LC and combining it with retrieval-augmented generation (#RAG) are explored, focusing on context length, relevance, and experiment design. ---- A summary generated by ChatGPT
#LLM course is finished with a successful #poster session. We had TEN projects & posters, with topics spanning from #foundation_model, LLM efficiency, #RAG, #longcontext LLM, to LLM for #coding, #IoT, #finance, etc. We thank the department @ND_CSE for supporting this event!
I'm off to EMNLP 2024 @ Miami to present LongAlign on long context alignment! Though LongAlign has been out for over six months, we're thrilled that our recent projects, LongWriter and LongCite, still build on its training pipelines. #EMNLP2024 #LongContext
๐ Our paper, "Multilingual Evaluation of Long Context Retrieval and Reasoning," has been accepted at MRL 2024, EMNLP. Our work sheds light on the challenges that long-context LLMs face in multilingual settings.
#emnlp2024 #nlp #multilingual #longcontext
We released the #Jamba 1.5 open model family:
- 256K #contextwindow
- Up to 2.5X faster on #longcontext in its size class
- Native support for structured JSON output, function calling, digesting doc objects & generating citations
https://t.co/tebBJW09c5
#AI #LLM #AI21Jamba
6/ โ ๏ธ Not Always Superior: LC LLMs don't always outshine RAG. For specific long datasets, like โBench's EnQA and EnMC, RAG surpasses LC models, especially when context lengths are extremely large (e.g., 147,000 words).
#AI #MachineLearning #RAG #LLM #LongContext #SELFROUTE
5/ ๐ฏ Versatility Across Tasks: The research evaluated the hybrid approach across various datasets, including QA tasks, multi-choice questions, and summarization, showcasing SELF-ROUTE's adaptability.
#AI #MachineLearning #RAG #LLM #LongContext #SELFROUTE
4/ ๐ค High Overlap in Predictions: Surprisingly, there's a 63% overlap in correct predictions between RAG and LC LLMs. This suggests a strong potential for hybrid models that maximize efficiency while maintaining accuracy.
#AI #MachineLearning #RAG #LLM #LongContext #SELFROUTE
InfiniGPT
----------
I have implemented the Infini Transformer, inspired by @Google & @CogArchitect's repo.
https://t.co/TxOlhPfPWU
#infiniattention #infinitransformer #research #generativeAI #gemini #googleresearch #llms #longcontext #GPT #llama #AI #openai #chatgpt4o #LLMs
LLM็ๆๆฐ็ญ็น่ฟๅฑ - ๅคงไธไธๆ! ่ฟๆฏๆจๅๆดๅนฟๆณๅบๆฏใๆด้ซ็บงๅบ็จๅ็ๆญฃๆบ่ฝ็ๆ ธๅฟๅ
ณ้ฎใ็ฎๅๅฐไบ่งฃไธไธ่ฟ๏ฟฝ๏ฟฝ๏ฟฝ้ฉๅฝๆง็ๆๆฏ๐:
๐ช RingAttention๏ผๅคงๅฐ1ไธชไบฟ
๐ฆ LongLLaMA๏ผๅผๆบ๏ผ256K
๐ Llama Long๏ผ32K็Metaๅฎๆน็ ็ฉถ
๐ง LongLoRA๏ผๅพฎ่ฐๆนๆณ๏ผๆฏๆๅฐ100K
#LLM #LongContext #GPT4 #GPT4V
Last Seen Hashtags on Sotwe
Most Popular Users
Elon Musk 
@elonmusk
240.2M followers
Barack Obama
@barackobama
119.3M followers
Donald J. Trump
@realdonaldtrump
111.6M followers
Cristiano Ronaldo
@cristiano
109.1M followers
Narendra Modi
@narendramodi
107M followers
Rihanna
@rihanna
97.3M followers
NASA
@nasa
92.1M followers
Justin Bieber
@justinbieber
90.6M followers
KATY PERRY
@katyperry
86.9M followers
Taylor Swift
@taylorswift13
80.7M followers
Lady Gaga
@ladygaga
72.3M followers
Kim Kardashian
@kimkardashian
69.4M followers
Virat Kohli
@imvkohli
68.7M followers
YouTube
@youtube
68.6M followers
Bill Gates
@billgates
63.5M followers
The Ellen Show
@theellenshow
62.5M followers
CNN
@cnn
61.9M followers
Neymar Jr
@neymarjr
61.3M followers
X
@x
60.9M followers
Selena Gomez
@selenagomez
60M followers