Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917745644273664 |
|---|---|
| author | Munkhdalai, Tsendsuren Faruqui, Manaal Gopal, Siddharth |
| author_facet | Munkhdalai, Tsendsuren Faruqui, Manaal Gopal, Siddharth |
| contents | This work introduces an efficient method to scale Transformer-based Large Language Models (LLMs) to infinitely long inputs with bounded memory and computation. A key component in our proposed approach is a new attention technique dubbed Infini-attention. The Infini-attention incorporates a compressive memory into the vanilla attention mechanism and builds in both masked local attention and long-term linear attention mechanisms in a single Transformer block. We demonstrate the effectiveness of our approach on long-context language modeling benchmarks, 1M sequence length passkey context block retrieval and 500K length book summarization tasks with 1B and 8B LLMs. Our approach introduces minimal bounded memory parameters and enables fast streaming inference for LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_07143 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention Munkhdalai, Tsendsuren Faruqui, Manaal Gopal, Siddharth Computation and Language Artificial Intelligence Machine Learning Neural and Evolutionary Computing This work introduces an efficient method to scale Transformer-based Large Language Models (LLMs) to infinitely long inputs with bounded memory and computation. A key component in our proposed approach is a new attention technique dubbed Infini-attention. The Infini-attention incorporates a compressive memory into the vanilla attention mechanism and builds in both masked local attention and long-term linear attention mechanisms in a single Transformer block. We demonstrate the effectiveness of our approach on long-context language modeling benchmarks, 1M sequence length passkey context block retrieval and 500K length book summarization tasks with 1B and 8B LLMs. Our approach introduces minimal bounded memory parameters and enables fast streaming inference for LLMs. |
| title | Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention |
| topic | Computation and Language Artificial Intelligence Machine Learning Neural and Evolutionary Computing |
| url | https://arxiv.org/abs/2404.07143 |