[serge] Fix 12 integration tests for model glm46v failing with other (other (12))

[sergereview]

Original CI failure

• Failure group: 1/1: 12 integration tests for model glm46vfailing withother (other (12))
• tests/models/glm46v/test_modeling_glm46v.py::Glm46VIntegrationTest::test_small_model_integration_test [multi-gpu] (other, seen 7/7)

Relates to #46906

The glm46v integration tests fail with RuntimeError: Expected tensor for argument #1 'indices' to have one of the following scalar types: Long, Int; but got CUDABFloat16Type instead while checking arguments for nn.Embedding.

Root cause: input_ids reach the text embedding lookup as floating-point tensors (bfloat16/float16) instead of integer indices. Glm46VModel inherits its forward from Glm4vModel/Qwen2VLModel, and the glm46v-specific subclass does not guard the embedding input dtype.

Fix: override Glm46VModel.forward to cast input_ids to torch.long before delegating to the parent forward, ensuring the embedding layer receives valid index tensors.

This is the minimal library-code change; no expected output values need updating.

---

This change was produced automatically by serge from a CI failure report. The patch was generated by an LLM and applied by serge; review before merging.

serge v0.1.0 · model: moonshotai/Kimi-K2.7-Code · 26 LLM turns · 30 tool calls · 192.2s · 581100 in / 15584 out tokens

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: glm46v

[github-actions]

CI Dashboard: View test results in Grafana

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.