Qualcomm AI Engine Direct - Add llama io be quantized

- Add general function to tag io obtain/genetate quantized tensor
- Add quantizing io function to llama2.py

Author: Joey Tsai

backends/qualcomm/quantizer/custom_annotation.py

@@ -12,6 +12,8 @@
     QuantizationConfig,
 )
 from executorch.backends.qualcomm.quantizer.utils import QUANT_ANNOTATION_KEY
+from executorch.exir.dialects._ops import ops as exir_ops
+from executorch.extension.llm.export.builder import LLMEdgeManager
 from torch.ao.quantization.quantizer import (
     QuantizationAnnotation,
     SharedQuantizationSpec,
@@ -144,3 +146,35 @@ def annotate_matmul(node: Node, quantization_config: QuantizationConfig):
     for node in gm.graph.nodes:
         if node.op == "call_function" and node.target == torch.ops.aten.matmul.default:
             annotate_matmul(node, quantization_config_16a8w)
+
+
+def get_custom_quant_ios_dtype(
+    cache_shape: torch.Size,
+    node: torch.fx.Node,
+    kv_dtype=torch.uint8,
+    sharding_dtype=torch.uint16,
+):
+    """
+    This function is specific for llama inputs and outputs
+    """
+    if node.op == "placeholder" and "attention_sdpa_kv_cache_past_" in node.name:
+        return kv_dtype
+
+    # Tag index put node before copy node, because copy is a skipped node in qnn
+    if (
+        exir_ops.edge.aten.index_put.default == node.target
+        and node.meta["val"].shape == cache_shape
+    ):
+        return kv_dtype
+
+    # Tag sharding io
+    if exir_ops.edge.llama.fallback.default in [
+        u.target for u in list(node.users.keys())
+    ] + [node.target]:
+        return sharding_dtype
+
+    # Tag index op as quantized tensors. It is caused by sharding
+    if exir_ops.edge.aten.index.Tensor in [
+        u.target for u in list(node.users.keys())
+    ] + [node.target]:
+        return sharding_dtype

backends/qualcomm/utils/utils.py

@@ -71,6 +71,7 @@
     QCOM_PASS_EXPAND_BROADCAST_SHAPE,
     QCOM_PASS_SKIP_ADVANCED_REQUANT,
     QCOM_QNN_COMPILE_SPEC,
+    QCOM_QUANTIZED_IO,
 )
 
 from executorch.exir import ExirExportedProgram
@@ -876,3 +877,12 @@ def get_soc_to_chipset_map():
         "SM8475": QcomChipset.SM8475,
         "SM8450": QcomChipset.SM8450,
     }
+
+
+def tag_quant_io(gm: torch.fx.GraphModule, get_quant_io_dtype_fn: Callable):
+    """
+    Tag io nodes which get/output quantized tensor. No need to insert q/dq in qnn_preprocess
+    """
+    for node in gm.graph.nodes:
+        if dtype := get_quant_io_dtype_fn(node):
+            node.meta[QCOM_QUANTIZED_IO] = dtype

examples/models/llama/export_llama_lib.py

@@ -643,8 +643,7 @@ def _export_llama(modelname, args) -> LLMEdgeManager:  # noqa: C901
             )
         )
         # pyre-ignore: Undefined import [21]: Could not find a module corresponding to import `executorch.backends.qualcomm.utils.utils`
-        from executorch.backends.qualcomm.utils.utils import _transform
-
+        from executorch.backends.qualcomm.utils.utils import _transform, tag_quant_io
         # pyre-ignore: Undefined attribute [16]: Module `executorch.backends` has no attribute `qualcomm`, Optional type has no attribute `exported_program`
         _transform(builder_exported_to_edge.edge_manager.exported_program())
 
@@ -656,6 +655,22 @@ def _export_llama(modelname, args) -> LLMEdgeManager:  # noqa: C901
                 shares=args.num_sharding,
             )
 
+        from functools import partial
+
+        from executorch.backends.qualcomm.quantizer.custom_annotation import (
+            get_custom_quant_ios_dtype,
+        )
+
+        tag_quant_io(
+            builder_exported_to_edge.edge_manager.exported_program().graph_module,
+            partial(
+                get_custom_quant_ios_dtype,
+                builder_exported_to_edge.model.layers[
+                    0
+                ].attention.kv_cache.past_k_caches.shape,
+            ),
+        )
+
     logging.info("Lowering model using following partitioner(s): ")
     for partitioner in partitioners:
         logging.info(f"--> {partitioner.__class__.__name__}")