Bug fix to add missing BRIDGE_TOWER_EMBEDDING env var for MultimodalQnA

[dmsuehir]

Description

There was a recent change to combine retriever microservices in GenAIComps. There was a coordinating PR updating MultimodalQnA, however it was missing these BRIDGE_TOWER_EMBEDDING environment variables in the Intel cpu/hpu README.md and set_vars.sh files.

Issues

Without setting this env var, querying the megaservice fails. The error in the lvm-llava-svc logs are:

File "/home/user/comps/lvms/llava/lvm.py", line 57, in lvm
    img_b64_str = retrieved_metadatas[0]["b64_img_str"]
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
KeyError: 'b64_img_str'

Type of change

List the type of change like below. Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

Dependencies

None

Tests

Manually tested, because the test scripts set the env vars as part of the test script, and those scripts were correctly setting the env var.

[github-actions]

Dependency Review

✅ No vulnerabilities or license issues found.

Scanned Files

[ashahba]

LGTM!