Access millions of research papers and academic content with full-text search:
Copy
response = valyu.search( "Extending context window of large language models via positional interpolation", search_type="proprietary", max_num_results=5, max_price=30, included_sources=["valyu/valyu-arxiv"])# Get detailed academic content with citationsprint(response.results[0].content)
Response includes rich academic content:
Copy
{ "success": true, "error": "", "tx_id": "tx_55e65c6f-3607-4ebe-892b-e964b9c72a8d", "query": "Extending context window of large language models via positional interpolation", "results": [ { "id": "55e65c6f-3607-4ebe-892b-e964b9c72a8d:2306.15595:1", "title": "Extending Context Window of Large Language Models via Positional Interpolation", "url": "https://arxiv.org/abs/2306.15595?utm_source=valyu.network&utm_medium=referral", "content": "#### 2.3 PROPOSED APPROACH: POSITION INTERPOLATION (PI)\n\n#### 2.1 BACKGROUND: ROTARY POSITION EMBEDDING (ROPE)\n\nTransformer models require explicit positional information to be injected, typically in the form of positional encodings, to represent the order of inputs. We consider Rotary Position Embedding (RoPE) (Su et al., 2021), which is the position encoding used in the LLaMA model...\n\n$$\\mathbf{f}(\\mathbf{x},m)=[(x_{0}+\\mathrm{i}x_{1})e^{im\\theta_{0}},(x_{2}+\\mathrm{i}x_{3})e^{im\\theta_{1}},\\ldots,(x_{d-2}+\\mathrm{i}x_{d-1})e^{im\\theta_{d/2-1}}]^{\\top}$$", "source": "valyu/valyu-arxiv", "length": 593, "image_url": { "_page_4_Figure_2.jpeg": "https://prod-s3-vyplatform-processeddata.s3.amazonaws.com/valyu/Arxiv-new/eb030308-a71b-5227-8ddf-9dbeeb1c6e12/_page_4_Figure_2.jpeg" }, "publication_date": "2023-01-01", "doi": "https://doi.org/10.48550/arxiv.2306.15595", "citation": "Shouyuan Chen et al. (2023). Extending Context Window of Large Language Models via Positional Interpolation.", "citation_count": 25, "authors": [ "Shouyuan Chen", "S.H. Wong", "Liangjian Chen", "Yuandong Tian" ], "price": 0.0005, "data_type": "unstructured", "source_type": "paper", "relevance_score": 0.8071867796187081 } ], "results_by_source": { "proprietary": 1, "web": 0 }}
Want financial data? Just ask in plain English - we’ll figure it out:
Copy
response = valyu.search( "Pfizer stock price since COVID-19 outbreak", search_type="proprietary", max_num_results=1, max_price=30 # Maximum price for a thousand retreivals (CPM))# The response includes structured JSON dataprint(response.results[0].content)
Our natural language processing is so good, it even understands queries like
”$$$$$ of larry and Sergey brins companie. on fr week commencing 5th hune on
the 21st century” 🤯 (Yes, that means “Google stock price for Friday June 5th,
2021” - we’re that good!)