arize-phoenix 0.0.30__py3-none-any.whl → 0.0.31rc2__py3-none-any.whl

{arize_phoenix-0.0.30.dist-info → arize_phoenix-0.0.31rc2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: arize-phoenix
-Version: 0.0.30
+Version: 0.0.31rc2
 Summary: ML Observability in your notebook
 Project-URL: Documentation, https://docs.arize.com/phoenix/
 Project-URL: Issues, https://github.com/Arize-ai/phoenix/issues
@@ -41,6 +41,9 @@ Requires-Dist: pytest; extra == 'dev'
 Requires-Dist: pytest-cov; extra == 'dev'
 Requires-Dist: pytest-lazy-fixture; extra == 'dev'
 Requires-Dist: strawberry-graphql[debug-server]==0.178.0; extra == 'dev'
+Provides-Extra: experimental
+Requires-Dist: openai; extra == 'experimental'
+Requires-Dist: tenacity; extra == 'experimental'
 Description-Content-Type: text/markdown
 
 <p align="center">

{arize_phoenix-0.0.30.dist-info → arize_phoenix-0.0.31rc2.dist-info}/RECORD

@@ -1,4 +1,4 @@
-phoenix/__init__.py,sha256=2qq5YT2m8nKOG5SGFwBiQNE-v7748SVHl9VCK6stMhc,1114
+phoenix/__init__.py,sha256=sfBQ2lmi-mb29hU-W-ZqstA9EALGX5n--2dKGf1PSJE,1192
 phoenix/config.py,sha256=tjNn9oqDxQmeO85sCchLlTsDiRJ6AoK0CTt_Uc_hrKM,1442
 phoenix/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
 phoenix/services.py,sha256=BlQF7lTQwhXUHBZBjZOoTnDM2Qni_hziUXsfp1Hux4Q,3978
@@ -15,6 +15,9 @@ phoenix/datasets/errors.py,sha256=-Iyk8rsvP_KX-P4gOqjm26slkDq1-9CohK07_LkrYCI,81
 phoenix/datasets/fixtures.py,sha256=0_PacL3dw49zulKpFpPdhvxJxeGmHTguqIyf2VXkBkk,19158
 phoenix/datasets/schema.py,sha256=HlM0f-pLFul2sYyHZM-Av8OFxLFkn57dkK_BWbMzyJY,6668
 phoenix/datasets/validation.py,sha256=dZ9lCFUV0EY7HCkQkQBrs-GLAEIZdpOqUxwD5l4dp88,8294
+phoenix/experimental/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+phoenix/experimental/evals/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+phoenix/experimental/evals/retrievals.py,sha256=3PBv2yFIcxdn_y-L6We554TsozqCDjTVBE8017RB2mY,3793
 phoenix/metrics/README.md,sha256=5gekqTU-5gGdMwvcfNp2Wlu8p1ul9kGY_jq0XXQusoI,1964
 phoenix/metrics/__init__.py,sha256=sLp7td1GIt_0Z8dPUyP4L0-_4x9c871yAaGX30oMsvg,2433
 phoenix/metrics/binning.py,sha256=CXPPcAkRmmR__IG36a6UGs5RBtgXXPuWQbafPtuG1ww,12787
@@ -84,12 +87,12 @@ phoenix/server/static/apple-touch-icon.png,sha256=fOfpjqGpWYbJ0eAurKsyoZP1EAs6ZV
 phoenix/server/static/favicon.ico,sha256=bY0vvCKRftemZfPShwZtE93DiiQdaYaozkPGwNFr6H8,34494
 phoenix/server/static/index.css,sha256=KKGpx4iwF91VGRm0YN-4cn8oC-oIqC6HecoPf0x3ZM8,1885
 phoenix/server/static/index.html,sha256=xPZZH-y4dWlbDutPEV1k0rhmWJtIV-Db9aYP-dEc7wM,703
-phoenix/server/static/index.js,sha256=Z7vMbIgbNLvLrx-FNUc84R8qvqfFm7XhQ2fCKZ3LvLU,2534545
+phoenix/server/static/index.js,sha256=SJnIFu7ufB_k38YBCI7D1btTSq4mmE7WkUf1iOBxiGw,2573278
 phoenix/server/static/modernizr.js,sha256=mvK-XtkNqjOral-QvzoqsyOMECXIMu5BQwSVN_wcU9c,2564
 phoenix/session/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/session/session.py,sha256=WwGH7qStR4kAhmrRsw35BwXvCQDnEWpGm0crjyrWTvs,9519
-arize_phoenix-0.0.30.dist-info/METADATA,sha256=zvzU09aJcBiVSBcrxJnY0OBRX6UBq_gyoshQtEp4S3s,10849
-arize_phoenix-0.0.30.dist-info/WHEEL,sha256=9QBuHhg6FNW7lppboF2vKVbCGTVzsFykgRQjjlajrhA,87
-arize_phoenix-0.0.30.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
-arize_phoenix-0.0.30.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
-arize_phoenix-0.0.30.dist-info/RECORD,,
+arize_phoenix-0.0.31rc2.dist-info/METADATA,sha256=iRpUMDo-rwHxcjDEUL4JY3HRi0Zma4Yz9nNWu0Z22hI,10977
+arize_phoenix-0.0.31rc2.dist-info/WHEEL,sha256=hKi7AIIx6qfnsRbr087vpeJnrVUuDokDHZacPPMW7-Y,87
+arize_phoenix-0.0.31rc2.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
+arize_phoenix-0.0.31rc2.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
+arize_phoenix-0.0.31rc2.dist-info/RECORD,,

{arize_phoenix-0.0.30.dist-info → arize_phoenix-0.0.31rc2.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: hatchling 1.18.0
+Generator: hatchling 1.12.2
 Root-Is-Purelib: true
 Tag: py3-none-any

phoenix/__init__.py

@@ -2,8 +2,9 @@ from .datasets.dataset import Dataset
 from .datasets.fixtures import ExampleDatasets, load_example
 from .datasets.schema import EmbeddingColumnNames, RetrievalEmbeddingColumnNames, Schema
 from .session.session import Session, active_session, close_app, launch_app
+from .trace.fixtures import load_example_traces
 
-__version__ = "0.0.30"
+__version__ = "0.0.31rc2"
 
 # module level doc-string
 __doc__ = """
@@ -32,4 +33,5 @@ __all__ = [
     "close_app",
     "launch_app",
     "Session",
+    "load_example_traces",
 ]

phoenix/experimental/evals/retrievals.py

@@ -0,0 +1,91 @@
+"""
+Helper functions for evaluating the retrieval step of retrieval-augmented generation.
+"""
+
+from typing import List, Optional
+
+import openai
+from tenacity import (
+    retry,
+    stop_after_attempt,
+    wait_random_exponential,
+)
+
+_EVALUATION_SYSTEM_MESSAGE = (
+    "You will be given a query and a reference text. "
+    "You must determine whether the reference text contains an answer to the input query. "
+    'Your response must be single word, either "relevant" or "irrelevant", '
+    "and should not contain any text or characters aside from that word. "
+    '"irrelevant" means that the reference text does not contain an answer to the query. '
+    '"relevant" means the reference text contains an answer to the query.'
+)
+_QUERY_CONTEXT_PROMPT_TEMPLATE = """# Query: {query}
+
+# Reference: {reference}
+
+# Answer ("relevant" or "irrelevant"): """
+
+
+def compute_precisions_at_k(
+    relevance_classifications: List[Optional[bool]],
+) -> List[Optional[float]]:
+    """Given a list of relevance classifications, computes precision@k for k = 1, 2, ..., n, where
+    n is the length of the input list.
+
+    Args:
+        relevance_classifications (List[Optional[bool]]): A list of relevance classifications for a
+            set of retrieved documents, sorted by order of retrieval (i.e., the first element is the
+            classification for the first retrieved document, the second element is the
+            classification for the second retrieved document, etc.). The list may contain None
+            values, which indicate that the relevance classification for the corresponding document
+            is unknown.
+
+    Returns:
+        List[Optional[float]]: A list of precision@k values for k = 1, 2, ..., n, where n is the
+            length of the input list. The first element is the precision@1 value, the second element
+            is the precision@2 value, etc. If the input list contains any None values, those values
+            are omitted when computing the precision@k values.
+    """
+    precisions_at_k = []
+    num_relevant_classifications = 0
+    num_non_none_classifications = 0
+    for relevance_classification in relevance_classifications:
+        if isinstance(relevance_classification, bool):
+            num_non_none_classifications += 1
+            num_relevant_classifications += int(relevance_classification)
+        precisions_at_k.append(
+            num_relevant_classifications / num_non_none_classifications
+            if num_non_none_classifications > 0
+            else None
+        )
+    return precisions_at_k
+
+
+@retry(wait=wait_random_exponential(min=1, max=60), stop=stop_after_attempt(6))
+def classify_relevance(query: str, document: str, model_name: str) -> Optional[bool]:
+    """Given a query and a document, determines whether the document contains an answer to the
+    query.
+
+    Args:
+        query (str): The query text. document (str): The document text. model_name (str): The name
+        of the OpenAI API model to use for the classification.
+
+    Returns:
+        Optional[bool]: A boolean indicating whether the document contains an answer to the query
+            (True meaning relevant, False meaning irrelevant), or None if the LLM produces an
+            unparseable output.
+    """
+    prompt = _QUERY_CONTEXT_PROMPT_TEMPLATE.format(
+        query=query,
+        reference=document,
+    )
+    response = openai.ChatCompletion.create(
+        messages=[
+            {"role": "system", "content": _EVALUATION_SYSTEM_MESSAGE},
+            {"role": "user", "content": prompt},
+        ],
+        model=model_name,
+    )
+    raw_response_text = str(response["choices"][0]["message"]["content"]).strip()
+    relevance_classification = {"relevant": True, "irrelevant": False}.get(raw_response_text)
+    return relevance_classification