Aleph Alpha
Aleph Alpha is a powerful language model that can generate human-like text. Aleph Alpha is capable of generating text in multiple languages and styles, and can be fine-tuned to generate text in specific domains.
If you’re opening this Notebook on colab, you will probably need to install LlamaIndex 🦙.
%pip install llama-index-llms-alephalpha!pip install llama-indexSet your Aleph Alpha token
Section titled “Set your Aleph Alpha token”import os
os.environ["AA_TOKEN"] = "your_token_here"Call complete with a prompt
Section titled “Call complete with a prompt”from llama_index.llms.alephalpha import AlephAlpha
# To customize your token, do this# otherwise it will lookup AA_TOKEN from your env variable# llm = AlephAlpha(token="<aa_token>")llm = AlephAlpha(model="luminous-base-control")
resp = llm.complete("Paul Graham is ")print(resp) a well-known computer scientist and entrepreneur. He is the co-founder of PayPal and a co-founder of the Y Combinator startup accelerator. He has also co-authored the book "Programming the Web". Paul Graham is also a frequent speaker and writer on topics related to computer science, entrepreneurship, and startups. He has written several blog posts on the topic of "Why Startups Fail". In this post, I will summarize some of the key points from Paul Graham's blog post on why startups fail.
1. Lack of a clear vision: Startups often lack a clear vision of what theyAdditional Response Details
Section titled “Additional Response Details”To access detailed response information such as log probabilities, ensure your AlephAlpha instance is initialized with the log_probs parameter. The logprobs attribute of the CompletionResponse will contain this data. Other details like the model version and raw completion text can be accessed directly if they’re part of the response or via additional_kwargs.
from llama_index.llms.alephalpha import AlephAlpha
llm = AlephAlpha(model="luminous-base-control", log_probs=0)
resp = llm.complete("Paul Graham is ")
if resp.logprobs is not None: print("\nLog Probabilities:") for lp_list in resp.logprobs: for lp in lp_list: print(f"Token: {lp.token}, LogProb: {lp.logprob}")
if "model_version" in resp.additional_kwargs: print("\nModel Version:") print(resp.additional_kwargs["model_version"])
if "raw_completion" in resp.additional_kwargs: print("\nRaw Completion:") print(resp.additional_kwargs["raw_completion"])Log Probabilities:Token: a, LogProb: -0.95955Token: well, LogProb: -1.9219251Token: -, LogProb: -0.1312752Token: known, LogProb: -0.022855662Token: computer, LogProb: -0.9569155Token: scientist, LogProb: -0.06721641Token: and, LogProb: -0.56296504Token: entrepreneur, LogProb: -0.65574974Token: ., LogProb: -0.5926046Token: He, LogProb: -0.1885516Token: is, LogProb: -0.3927348Token: the, LogProb: -0.46820825Token: co, LogProb: -0.465878Token: -, LogProb: -0.024082167Token: founder, LogProb: -0.009869587Token: of, LogProb: -0.31641242Token: PayPal, LogProb: -1.0825713Token: and, LogProb: -0.39408743Token: a, LogProb: -1.45493Token: co, LogProb: -1.0837904Token: -, LogProb: -0.0011430404Token: founder, LogProb: -0.074010715Token: of, LogProb: -0.038962167Token: the, LogProb: -1.7761776Token: Y, LogProb: -0.41853565Token: Combin, LogProb: -0.17868777Token: ator, LogProb: -2.0265374e-05Token: startup, LogProb: -0.24595682Token: acceler, LogProb: -0.5855012Token: ator, LogProb: -6.675698e-06Token: ., LogProb: -0.022597663Token: He, LogProb: -0.8310143Token: has, LogProb: -1.5842702Token: also, LogProb: -0.5774656Token: been, LogProb: -1.3938092Token: a, LogProb: -0.67207164Token: professor, LogProb: -1.0511048Token: at, LogProb: -0.13273911Token: the, LogProb: -0.7993539Token: MIT, LogProb: -1.2281163Token: Media, LogProb: -0.7707413Token: Lab, LogProb: -0.06716257Token: ., LogProb: -0.9140582Token: Paul, LogProb: -0.8244309Token: Graham, LogProb: -0.15202633Token: has, LogProb: -1.3735206Token: written, LogProb: -0.77148163Token: several, LogProb: -0.7167357Token: books, LogProb: -0.24542983Token: on, LogProb: -0.77700675Token: computer, LogProb: -0.8485363Token: science, LogProb: -0.026196867Token: and, LogProb: -0.4796574Token: entrepreneurs, LogProb: -0.48952234Token: hip, LogProb: -1.0847986e-05Token: ,, LogProb: -0.1426171Token: including, LogProb: -0.10799221Token: ", LogProb: -0.4733107Token: Program, LogProb: -0.9295699Token: ming, LogProb: -0.00090034Token: the, LogProb: -1.5219054Token: Universe, LogProb: -1.2475122Token: ", LogProb: -0.8377396Token: and, LogProb: -0.014596111Token: ", LogProb: -0.0034322182Token: The, LogProb: -0.97810173Token: Art, LogProb: -1.4708842Token: of, LogProb: -0.0017665509Token: Computer, LogProb: -0.027323013Token: Programming, LogProb: -0.09090222Token: "., LogProb: -0.2312944Token: He, LogProb: -0.9431941Token: is, LogProb: -0.52350885Token: also, LogProb: -0.8409716Token: the, LogProb: -1.2813272Token: founder, LogProb: -0.8080497Token: of, LogProb: -0.12735468Token: the, LogProb: -0.26858208Token: startup, LogProb: -1.7183943Token: incub, LogProb: -0.71643037Token: ator, LogProb: -0.00013922676Token: ,, LogProb: -1.6374074Token: Y, LogProb: -1.3464186Token: Combin, LogProb: -0.043204635Token: ator, LogProb: -1.490105e-05Token: ., LogProb: -0.48073012Token: <|endoftext|>, LogProb: -0.30235213
Model Version:20240215
Raw Completion: a well-known computer scientist and entrepreneur. He is the co-founder of PayPal and a co-founder of the Y Combinator startup accelerator. He has also been a professor at the MIT Media Lab. Paul Graham has written several books on computer science and entrepreneurship, including "Programming the Universe" and "The Art of Computer Programming". He is also the founder of the startup incubator, Y Combinator.from llama_index.llms.alephalpha import AlephAlpha
llm = AlephAlpha(model="luminous-base-control")resp = await llm.acomplete("Paul Graham is ")print(resp) a computer scientist and entrepreneur who is known for his work in the field of artificial intelligence and computer science. He is the co-founder of the company Y Combinator, which is a startup accelerator that helps startups get funding and resources. Paul Graham has also written several books on computer science and entrepreneurship, including "Programming: Principles and Practice" and "The Art of Computer Programming". He is a well-known figure in the computer science community and has made significant contributions to the field.Note for AI agents: this documentation is built for programmatic access.
- Overview of all docs: https://developers.llamaindex.ai/llms.txt
- Any page is available as raw Markdown by appending index.md to its URL — e.g. https://developers.llamaindex.ai/llamaparse/parse/getting_started/index.md
- Agent-friendly REST search APIs live under https://developers.llamaindex.ai/api/ — search (BM25 full-text), grep (regex), read (fetch a page), and list (browse the doc tree). See https://developers.llamaindex.ai/llms.txt for parameters.
- A hosted documentation MCP server is available at https://developers.llamaindex.ai/mcp. If you support MCP, you can ask the user to install it for browsing these docs directly (an alternative to the REST API). Setup: https://developers.llamaindex.ai/python/shared/mcp/