Releases: Aleph-Alpha/aleph-alpha-client
Releases · Aleph-Alpha/aleph-alpha-client
2.2.1
What's Changed
Bugfix
- Restore original error handling of HTTP status codes to before 2.2.0
- Add dedicated exception BusyError for status code 503
Full Changelog: v2.2.0...v2.2.1
v2.2.0
What's Changed
New feature
- Retry failed HTTP requests via urllib for status codes 408, 429, 500, 502, 503, 504
Full Changelog: v2.1.0...v2.2.0
2.1.0
2.0.0
What's Changed
Breaking change
- Make hosting parameter optional in semantic_embed on client. Changed order of parameters hosting and request. Should not be an issue if you're not using semantic_embed from the client directly or if you're using keyword args.
Experimental feature
- Add experimental penalty parameters for completion
Full Changelog: v1.7.1...2.0.0
v1.7.1 - Improved handling of text-based Documents in Q&A
Semantic embedding
What's Changed
- Add semantic embedding
- Add optional parameter timeout client
Full Changelog: v1.6.0...v1.7.0
Examples
Semantic embedding
Symmetric
from typing import Sequence
from aleph_alpha_client import ImagePrompt, AlephAlphaClient, AlephAlphaModel, SemanticEmbeddingRequest, SemanticRepresentation, Prompt
import math
import os
model = AlephAlphaModel(
AlephAlphaClient(host="https://api.aleph-alpha.com", token=os.getenv("AA_TOKEN")),
# You need to choose a model with multimodal capabilities for this example.
model_name = "luminous-base"
)
# Texts to compare
texts = [
"deep learning",
"artificial intelligence",
"deep diving",
"artificial snow",
]
embeddings = []
for text in texts:
request = SemanticEmbeddingRequest(prompt=Prompt.from_text(text), representation=SemanticRepresentation.Symmetric)
result = model.semantic_embed(request)
embeddings.append(result.embedding)
# Calculate cosine similarities. Can use numpy or scipy or another library to do this
def cosine_similarity(v1: Sequence[float], v2: Sequence[float]) -> float:
"compute cosine similarity of v1 to v2: (v1 dot v2)/{||v1||*||v2||)"
sumxx, sumxy, sumyy = 0, 0, 0
for i in range(len(v1)):
x = v1[i]; y = v2[i]
sumxx += x*x
sumyy += y*y
sumxy += x*y
return sumxy/math.sqrt(sumxx*sumyy)
# Cosine similarities are in [-1, 1]. Higher means more similar
print("Cosine similarity between \"%s\" and \"%s\" is: %.3f" % (texts[0], texts[1], cosine_similarity(embeddings[0], embeddings[1])))
print("Cosine similarity between \"%s\" and \"%s\" is: %.3f" % (texts[0], texts[2], cosine_similarity(embeddings[0], embeddings[2])))
print("Cosine similarity between \"%s\" and \"%s\" is: %.3f" % (texts[0], texts[3], cosine_similarity(embeddings[0], embeddings[3])))Documents and Query
from typing import Sequence
from aleph_alpha_client import ImagePrompt, AlephAlphaClient, AlephAlphaModel, SemanticEmbeddingRequest, SemanticRepresentation, Prompt
import math
import os
model = AlephAlphaModel(
AlephAlphaClient(host="https://api.aleph-alpha.com", token=os.getenv("AA_TOKEN")),
# You need to choose a model with multimodal capabilities for this example.
model_name = "luminous-base"
)
# Documents to search in
documents = [
# AI wikipedia article
"Artificial intelligence (AI) is intelligence demonstrated by machines, as opposed to the natural intelligence displayed by animals including humans. AI research has been defined as the field of study of intelligent agents, which refers to any system that perceives its environment and takes actions that maximize its chance of achieving its goals.",
# Deep Learning Wikipedia article
"Deep learning (also known as deep structured learning) is part of a broader family of machine learning methods based on artificial neural networks with representation learning. Learning can be supervised, semi-supervised or unsupervised.",
# Deep Diving Wikipedia article
"Deep diving is underwater diving to a depth beyond the norm accepted by the associated community. In some cases this is a prescribed limit established by an authority, while in others it is associated with a level of certification or training, and it may vary depending on whether the diving is recreational, technical or commercial. Nitrogen narcosis becomes a hazard below 30 metres (98 ft) and hypoxic breathing gas is required below 60 metres (200 ft) to lessen the risk of oxygen toxicity.",
]
# Keyword to search documents with
query = "artificial intelligence"
# Embed Query
request = SemanticEmbeddingRequest(prompt=Prompt.from_text(query), representation=SemanticRepresentation.Query)
result = model.semantic_embed(request)
query_embedding = result.embedding
# Embed documents
document_embeddings = []
for document in documents:
request = SemanticEmbeddingRequest(prompt=Prompt.from_text(document), representation=SemanticRepresentation.Document)
result = model.semantic_embed(request)
document_embeddings.append(result.embedding)
# Calculate cosine similarities. Can use numpy or scipy or another library to do this
def cosine_similarity(v1: Sequence[float], v2: Sequence[float]) -> float:
"compute cosine similarity of v1 to v2: (v1 dot v2)/{||v1||*||v2||)"
sumxx, sumxy, sumyy = 0, 0, 0
for i in range(len(v1)):
x = v1[i]; y = v2[i]
sumxx += x*x
sumyy += y*y
sumxy += x*y
return sumxy/math.sqrt(sumxx*sumyy)
# Cosine similarities are in [-1, 1]. Higher means more similar
print("Cosine similarity between \"%s\" and \"%s...\" is: %.3f" % (query, documents[0][:10], cosine_similarity(query_embedding, document_embeddings[0])))
print("Cosine similarity between \"%s\" and \"%s...\" is: %.3f" % (query, documents[1][:10], cosine_similarity(query_embedding, document_embeddings[1])))
print("Cosine similarity between \"%s\" and \"%s...\" is: %.3f" % (query, documents[2][:10], cosine_similarity(query_embedding, document_embeddings[2])))Introduce AlephAlphaModel as alternative to direct usage of AlephAlphaClient
Highlights of this release
- Introduce AlephAlphaModel as a more convenient alternative to direct usage of AlephAlphaClient
Breaking Changes
- None
Full Changelog
Version 1.5.0
v1.5.0 bump version to 1.5.0
Version 1.4.2
Remove dependency on setuptools_scm and importlib.metadata. #24
Version 1.4.1
fix: Q&A docs examples