KeyError: 'tokens'

[Anton293]

The model expects embeddings as input.

# pip install git+https://github.com/google-ai-edge/ai-edge-apis.git#subdirectory=litert_tools

from litert_tools.pipeline import pipeline
runner = pipeline.load("gemma-3n-E4B-it-int4.task")

prompt = "hello world! "
output = runner.generate(prompt, max_decode_steps=500)