[Executorch][llama] Update SDPA op to use quantized kv cache

[kimishpatel]

Stack from ghstack (oldest at bottom):

• [Executorch][quant] Optimize per channel dequantize #5656
• [Executorch][llama] Add custom_sdpa and use that instead of sdpa_with_kv_cache #5655
• [Executorch] Update EXECUTORCH_LIBRARY macro #5654
• [Executorch][llama] Refactoring sdpa #5653
• -> [Executorch][llama] Update SDPA op to use quantized kv cache #5652
• Add update_quantized_cache op #5527
• Refactor custom SDPA op to separate kv cache update from the custom sdpa op #5651
• [ExecuTorch] Add quantized kv cache to llama #5650
• Fix dequantize per channel to handle double scale type #5524
• [ExecuTorch] Some updated to kv cache #5649

Using quantized kv cache, we cannot rely on sdpa to update the original case.
SO we insert cache update op

Differential Revision: D62301841

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/5652

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 6 New Failures

As of commit 85b2304 with merge base d2ba238 ():

NEW FAILURES - The following jobs have failed:

• pull / test-llama-runner-linux (bf16, buck2, portable) / linux-job (gh)
  RuntimeError: Command docker exec -t 7b08f0074d8d3cf3a1d24148f4e3a82caa29fb68fbcde8e84d858bd5a58d8301 /exec failed with exit code 3
• pull / test-llama-runner-linux (fp32, buck2, portable) / linux-job (gh)
  RuntimeError: Command docker exec -t 6b8d6ce8292d69362fd453262550ddf431246deb1e16cca01cdc02eedcd6378a /exec failed with exit code 3
• pull / test-llama-runner-linux (fp32, buck2, xnnpack+custom) / linux-job (gh)
  RuntimeError: Command docker exec -t c8fc08a2a9a8befcf04d370e8a2c31f474b64436739b59b044dd28be4cfa1821 /exec failed with exit code 3
• pull / test-llama-runner-linux (fp32, buck2, xnnpack+custom+qe) / linux-job (gh)
  RuntimeError: Command docker exec -t ce5623cfc3b78c718479e6795f707b22bd0e12e5ac8d5876da0d05fc0eebb536 /exec failed with exit code 3
• pull / unittest / linux / linux-job (gh)
  kernels/quantized/test/test_quant_dequant_per_token.py::QuantizePerTokenTest::test_quantize_per_token_large_tensor
• pull / unittest / macos / macos-job (gh)
  kernels/quantized/test/test_quant_dequant_per_token.py::QuantizePerTokenTest::test_quantize_per_token_large_tensor

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D62301841

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D62301841