UPSTREAM PR #18658: model-conversion : add detect_pooling script

[loci-dev]

Mirrored from ggml-org/llama.cpp#18658

This commit adds a Python script to automatically detect the pooling configuration from a sentence-transformers model directory.

The motivation for this change is that I make a mistake when adding the sentence-transformers support and I incorrectly assumed that if an embedding model uses sentence-transformers, it always used pooling. With the recent addition of support for late interaction models, which can have a down-projection but do not use pooling (like LFM2-ColBert-350M).

This commit builds upon ggml-org/llama.cpp#18464 which needs to be merged first.

Refs: ggml-org/llama.cpp#18607 (comment)

[loci-review]

Explore the complete analysis inside the Version Insights

Perfect! I've retrieved the summary report for your project. Here's what the analysis shows:

Summary Report for llama.cpp PR #843

Key Findings:

• ✅ No significant performance changes detected
• The analysis compared two versions and found no modified functions with performance changes greater than 2% in either response time or throughput time
• This indicates a neutral performance impact - the changes maintain performance stability

Project Information:

• Repository: llama.cpp (auroralabs-loci)
• Pull Request: UPSTREAM PR #18658: model-conversion : add detect_pooling script #843
• Project ID: 2621b8c0-b5ce-11f0-b333-453f42058aa1
• Report ID: 7956da10-ebad-11f0-81f2-dbb430499cb5

This is a positive result indicating that your code changes don't introduce any performance regressions or unexpected improvements, maintaining consistent performance characteristics across the codebase.

Would you like more detailed information about specific functions or any other aspect of this performance comparison?