Add GPT4Vision Image loader (#1089)

Co-authored-by: Deshraj Yadav <deshrajdry@gmail.com>
This commit is contained in:
Sidharth Mohanty
2024-01-02 03:57:23 +05:30
committed by GitHub
parent 367d6b70e2
commit c62663f2e4
29 changed files with 291 additions and 714 deletions

View File

@@ -148,73 +148,6 @@ def test_chroma_db_collection_changes_encapsulated():
app.db.reset()
def test_chroma_db_collection_add_with_skip_embedding(app_with_settings):
# Start with a clean app
app_with_settings.db.reset()
assert app_with_settings.db.count() == 0
app_with_settings.db.add(
embeddings=[[0, 0, 0]],
documents=["document"],
metadatas=[{"url": "url_1", "doc_id": "doc_id_1"}],
ids=["id"],
skip_embedding=True,
)
assert app_with_settings.db.count() == 1
data = app_with_settings.db.get(["id"], limit=1)
expected_value = {
"documents": ["document"],
"embeddings": None,
"ids": ["id"],
"metadatas": [{"url": "url_1", "doc_id": "doc_id_1"}],
"data": None,
"uris": None,
}
assert data == expected_value
data_without_citations = app_with_settings.db.query(
input_query=[0, 0, 0], where={}, n_results=1, skip_embedding=True
)
expected_value_without_citations = ["document"]
assert data_without_citations == expected_value_without_citations
app_with_settings.db.reset()
def test_chroma_db_collection_add_with_invalid_inputs(app_with_settings):
# Start with a clean app
app_with_settings.db.reset()
assert app_with_settings.db.count() == 0
with pytest.raises(ValueError):
app_with_settings.db.add(
embeddings=[[0, 0, 0]],
documents=["document", "document2"],
metadatas=[{"value": "somevalue"}],
ids=["id"],
skip_embedding=True,
)
assert app_with_settings.db.count() == 0
with pytest.raises(ValueError):
app_with_settings.db.add(
embeddings=None,
documents=["document", "document2"],
metadatas=[{"value": "somevalue"}],
ids=["id"],
skip_embedding=True,
)
assert app_with_settings.db.count() == 0
app_with_settings.db.reset()
def test_chroma_db_collection_collections_are_persistent():
db = ChromaDB(config=ChromaDbConfig(allow_reset=True, dir="test-db"))
app = App(config=AppConfig(collect_metrics=False), db=db)
@@ -312,60 +245,3 @@ def test_chroma_db_collection_reset():
app2.db.reset()
app3.db.reset()
app4.db.reset()
def test_chroma_db_collection_query(app_with_settings):
app_with_settings.db.reset()
assert app_with_settings.db.count() == 0
app_with_settings.db.add(
embeddings=[[0, 0, 0]],
documents=["document"],
metadatas=[{"url": "url_1", "doc_id": "doc_id_1"}],
ids=["id"],
skip_embedding=True,
)
assert app_with_settings.db.count() == 1
app_with_settings.db.add(
embeddings=[[0, 1, 0]],
documents=["document2"],
metadatas=[{"url": "url_2", "doc_id": "doc_id_2"}],
ids=["id2"],
skip_embedding=True,
)
assert app_with_settings.db.count() == 2
data_without_citations = app_with_settings.db.query(
input_query=[0, 0, 0], where={}, n_results=2, skip_embedding=True
)
expected_value_without_citations = ["document", "document2"]
assert data_without_citations == expected_value_without_citations
data_with_citations = app_with_settings.db.query(
input_query=[0, 0, 0], where={}, n_results=2, skip_embedding=True, citations=True
)
expected_value_with_citations = [
(
"document",
{
"url": "url_1",
"doc_id": "doc_id_1",
"score": 0.0,
},
),
(
"document2",
{
"url": "url_2",
"doc_id": "doc_id_2",
"score": 1.0,
},
),
]
assert data_with_citations == expected_value_with_citations
app_with_settings.db.reset()