tico 0.1.0.dev250623__py3-none-any.whl → 0.1.0.dev250625__py3-none-any.whl

tico/__init__.py

@@ -21,7 +21,7 @@ from tico.config import CompileConfigV1, get_default_config
 from tico.utils.convert import convert, convert_from_exported_program, convert_from_pt2
 
 # THIS LINE IS AUTOMATICALLY GENERATED BY setup.py
-__version__ = "0.1.0.dev250623"
+__version__ = "0.1.0.dev250625"
 
 MINIMUM_SUPPORTED_VERSION = "2.5.0"
 SECURE_TORCH_VERSION = "2.6.0"

tico/experimental/quantization/passes/fold_quant_ops.py

@@ -21,9 +21,8 @@ import copy
 import torch
 from torch.export import ExportedProgram
 
-from tico.serialize.quant_param import QPARAM_KEY, QuantParam, to_qparam_dtype
+from tico.serialize.quant_param import QPARAM_KEY, QuantParam
 from tico.utils import logging
-from tico.utils.graph import create_node
 from tico.utils.passes import PassBase, PassResult
 from tico.utils.trace_decorators import trace_graph_diff_on_pass
 from tico.utils.utils import get_quant_dtype
@@ -42,11 +41,33 @@ class FoldQuantOps(PassBase):
     To export quantized circle, this pass removes (Q - DQ) nodes and saves those quantization info
      to previous op's metadata.
 
-    [BEFORE]
-      op (float) - Quantize - Dequantize - (float)
-
-    [AFTER]
-      op (float with meta[QPARAM_KEY])
+    ────────────────────────────────────────────────────────────────
+    BEFORE                             AFTER
+    ────────────────────────────────────────────────────────────────
+      op(float) ─ Q ─ DQ ─ …            op(float, meta[QPARAM])
+
+      op ─ Q1 ─ DQ1 ─ Q2 ─ DQ2          op(meta[QPARAM]) ─ Q2
+                 ▲                                          ▲
+                 │ (Q1, DQ1 folded)                         │ (re-quantization kept)
+
+      op ─ Q ─┬─ DQ0                    op(meta[QPARAM])
+              ├─ DQ1                    (each DQ* folded, Q dropped when orphaned)
+              └─ DQ2
+    ────────────────────────────────────────────────────────────────
+
+    Algorithm
+    ---------
+    1. Iterate over *all* Dequantize nodes.
+    2. For each DQ, verify it is driven by a Quantize node `q` and that
+       `q` and `dq` share identical (scale, zero-point, dtype).
+    3. a) If the producer op has **no** QPARAM, attach one, then replace
+          *this* DQ's usages with the producer op.
+       b) If the producer is already quantized with a different dtype,
+          this is a *re-quantization*: attach QPARAM to `q` and keep it,
+          but still remove the DQ.
+    4. After all replacements, run `graph.eliminate_dead_code()`.
+       Any Quantize that became orphaned because *all* its DQs were folded
+       is deleted automatically.
     """
 
     def __init__(self):
@@ -81,15 +102,9 @@ class FoldQuantOps(PassBase):
             if q_args.dtype != dq_args.dtype:
                 continue
 
-            # Case 1. op is not quantized
-            # - Quantize op
-            # Case 2. op is quantized
-            # 2.1. op_dtype == qdq_dtype
-            # - Just skip (NOTE Need requantization?)
-            # 2.2. op_dtype != qdq_dtype
-            # - Insert Quantize operator
-
-            # Case 1
+            # ───────────────────────────────────────────
+            # Case 1: op not yet quantized
+            # ───────────────────────────────────────────
             if QPARAM_KEY not in op.meta:
                 qparam = QuantParam()
                 qparam.scale = [q_args.scale]
@@ -100,21 +115,36 @@ class FoldQuantOps(PassBase):
                 dq.replace_all_uses_with(op, propagate_meta=False)
 
                 logger.debug(f"{q.name} and {dq.name} are folded to {op.name}.")
+            # ───────────────────────────────────────────
+            # Case 2: op already quantized
+            #        2.1 same dtype  → nothing to do
+            #        2.2 diff dtype  → leave Q in place
+            # ───────────────────────────────────────────
             else:
                 op_qparam: QuantParam = op.meta[QPARAM_KEY]
                 qdq_dtype = get_quant_dtype(q_args.quant_min, q_args.quant_max)
-                # Case 2.2
+
                 if op_qparam.dtype != qdq_dtype:
-                    # If op is already quantized with a dtype different from qdq, leave quantize
-                    qparam = QuantParam()
-                    qparam.scale = [q_args.scale]
-                    qparam.zero_point = [q_args.zero_p]
-                    qparam.dtype = qdq_dtype
-                    q.meta[QPARAM_KEY] = qparam
-                    assert len(q.users) == 1, "Fix me unless"
+                    # Attach QPARAM to Q once
+                    if QPARAM_KEY not in q.meta:
+                        qparam = QuantParam()
+                        qparam.scale = [q_args.scale]
+                        qparam.zero_point = [q_args.zero_p]
+                        qparam.dtype = qdq_dtype
+                        q.meta[QPARAM_KEY] = qparam
+                        assert len(q.users) == 1, "Fix me unless"
 
                     dq.replace_all_uses_with(q, propagate_meta=False)
                     logger.debug(f"{dq.name} is folded ({q.name} is left).")
+                else:
+                    # Same dtype → the Quantize–Dequantize pair is redundant.
+                    assert op_qparam.scale and op_qparam.scale[0] == q_args.scale
+                    assert (
+                        op_qparam.zero_point
+                        and op_qparam.zero_point[0] == q_args.zero_p
+                    )
+                    dq.replace_all_uses_with(op, propagate_meta=False)
+                    logger.debug(f"Removed redundant {dq.name}")
 
         graph.eliminate_dead_code()
         graph.lint()

{tico-0.1.0.dev250623.dist-info → tico-0.1.0.dev250625.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: tico
-Version: 0.1.0.dev250623
+Version: 0.1.0.dev250625
 Summary: Convert exported Torch module to circle
 Home-page: UNKNOWN
 License: UNKNOWN

{tico-0.1.0.dev250623.dist-info → tico-0.1.0.dev250625.dist-info}/RECORD

@@ -1,4 +1,4 @@
-tico/__init__.py,sha256=nwI5rrfUKNWpensANra_9s0rBRWY4gFZmXqyKL0suYI,1743
+tico/__init__.py,sha256=doEv0YyghxyOT9H264hpfuazFicVamcLVaGtfVzXDNc,1743
 tico/pt2_to_circle.py,sha256=gu3MD4Iqc0zMZcCZ2IT8oGbyj21CTSbT3Rgd9s2B_9A,2767
 tico/config/__init__.py,sha256=xZzCXjZ84qE-CsBi-dfaL05bqpQ3stKKfTXhnrJRyVs,142
 tico/config/base.py,sha256=anwOiJFkUxUi7Cef573JgQcjk6S-FSi6O_TLjYASW-g,1244
@@ -50,7 +50,7 @@ tico/experimental/quantization/evaluation/executor/backend_executor.py,sha256=3k
 tico/experimental/quantization/evaluation/executor/circle_executor.py,sha256=eCCJ9wTwR0vUJ0oN7jxtQxZ9598GRw6P6KUxiuGsIIM,2685
 tico/experimental/quantization/evaluation/executor/triv24_executor.py,sha256=sUoXl6oOO2arAKaNjOBg7HiQja145_Jv6qgY7XtR7A8,5159
 tico/experimental/quantization/passes/__init__.py,sha256=IO6FP_xYbGy0dW0HL26GXD3ouxARaxCK7bz9dn4blPQ,26
-tico/experimental/quantization/passes/fold_quant_ops.py,sha256=cC5xuvqWqueKzLXUJi0pCk85vryImkhgnx-ee7G-tPI,4463
+tico/experimental/quantization/passes/fold_quant_ops.py,sha256=iaBMyO49CwVkhebMz3rjkHWfWE2LhwH6fORe7n4S6XQ,7040
 tico/experimental/quantization/passes/insert_quantize_on_dtype_mismatch.py,sha256=AbNcI7rfIwHsQna_rFuwqFdOzFAU2lIB3sMK-vns8Dc,13072
 tico/experimental/quantization/passes/propagate_qparam_backward.py,sha256=TGtyW0Z2qOTgVIasBdGRgbwH31YYd6ek7OvLTmCV614,3118
 tico/experimental/quantization/passes/propagate_qparam_forward.py,sha256=RhUHGCR2RpBO5KYkQ7Z8U5u7HEwDq2wdKHLKAJCi-5c,5138
@@ -198,9 +198,9 @@ tico/utils/mx/__init__.py,sha256=IO6FP_xYbGy0dW0HL26GXD3ouxARaxCK7bz9dn4blPQ,26
 tico/utils/mx/elemwise_ops.py,sha256=V6glyAHsVR1joqpsgnNytatCD_ew92xNWZ19UFDoMTA,10281
 tico/utils/mx/formats.py,sha256=uzNWyu-1onUlwQfX5cZ6fZSUfHMRqorper7_T1k3jfk,3404
 tico/utils/mx/mx_ops.py,sha256=RcfUTYVi-wilGB2sC35OeARdwDqnixv7dG5iyZ-fQT8,8555
-tico-0.1.0.dev250623.dist-info/LICENSE,sha256=kp4JLII7bzRhPb0CPD5XTDZMh22BQ7h3k3B7t8TiSbw,12644
-tico-0.1.0.dev250623.dist-info/METADATA,sha256=MtM7eJXlvpwfklWOfUefNC2IwQ2-YFkYUDGQRur0x6A,8846
-tico-0.1.0.dev250623.dist-info/WHEEL,sha256=G16H4A3IeoQmnOrYV4ueZGKSjhipXx8zc8nu9FGlvMA,92
-tico-0.1.0.dev250623.dist-info/entry_points.txt,sha256=kBKYSS_IYrSXmUYevmmepqIVPScq5vF8ulQRu3I_Zf0,59
-tico-0.1.0.dev250623.dist-info/top_level.txt,sha256=oqs7UPoNSKZEwqsX8B-KAWdQwfAa7i60pbxW_Jk7P3w,5
-tico-0.1.0.dev250623.dist-info/RECORD,,
+tico-0.1.0.dev250625.dist-info/LICENSE,sha256=kp4JLII7bzRhPb0CPD5XTDZMh22BQ7h3k3B7t8TiSbw,12644
+tico-0.1.0.dev250625.dist-info/METADATA,sha256=bCYr0NfPjTLo4lCwgJS_mwIMzDRJjpIUF9cRGrQRKOk,8846
+tico-0.1.0.dev250625.dist-info/WHEEL,sha256=G16H4A3IeoQmnOrYV4ueZGKSjhipXx8zc8nu9FGlvMA,92
+tico-0.1.0.dev250625.dist-info/entry_points.txt,sha256=kBKYSS_IYrSXmUYevmmepqIVPScq5vF8ulQRu3I_Zf0,59
+tico-0.1.0.dev250625.dist-info/top_level.txt,sha256=oqs7UPoNSKZEwqsX8B-KAWdQwfAa7i60pbxW_Jk7P3w,5
+tico-0.1.0.dev250625.dist-info/RECORD,,