[Devx:119] Models Interface

[sainivedh]

What

• Added functionality to interface with Models(predict, list, versions)

How

Model Predict

from clarifai.client.model import Model

# Model Predict
model = Model(user_id="user_id", app_id="app_id", model_id="model_id")
model_prediction = model.predict_by_url(url="url", input_type="image") # Supports image, text, audio, video

# Customizing Model Inference Output
model = Model(user_id="user_id", app_id="app_id", model_id="model_id",
                  output_config={"min_value": 0.98}) # Return predictions having prediction confidence > 0.98
model_prediction = model.predict_by_filepath(filepath="local_filepath", input_type="text") # Supports image, text, audio, video

model = Model(user_id="user_id", app_id="app_id", model_id="model_id",
                    output_config={"sample_ms": 2000}) # Return predictions for specified interval
model_prediction = model.predict_by_url(url=BEER_VIDEO_URL, input_type="video")

Models Listing

# List all model versions
all_model_versions = model.list_versions()

# Go to specific model version
model_v1 = client.app("app_id").model("model_id", model_version_id="model_version_id")

# List all models in an app
all_models = app.list_models()

# List all models in community filtered by model_type, description
all_llm_community_models = App().list_models(filter_by={"query": "LLM",
                                                        "model_type_id": "text-to-text"}, only_in_app=False)

Tests

Unit tests added for Model predict with different scenarios

• Predict by filepath, url, bytes
• Predict by various output configs (max_concepts, min_value, selected_concepts)

Note

Skipping Model unit tests in build as secrets needed for PAT are not yet setup for this repo

[stmugisha]

Looking good.
just a few comments

[sanjaychelliah]

👍

[stmugisha]

LGTM