ci: update version string in docs

docs/python/requirements-docs.txt

@@ -1,4 +1,4 @@
-ezkl==0.0.0
+ezkl==15.1.2
 sphinx
 sphinx-rtd-theme
 sphinxcontrib-napoleon

docs/python/src/conf.py

@@ -1,7 +1,7 @@
 import ezkl
 
 project = 'ezkl'
-release = '0.0.0'
+release = '15.1.2'
 version = release
 
 

examples/onnx/rsqrt/gen.py

@@ -0,0 +1,42 @@
+from torch import nn
+import torch
+import json
+import numpy as np
+
+
+class MyModel(nn.Module):
+    def __init__(self):
+        super(MyModel, self).__init__()
+
+    def forward(self, x):         
+        # reciprocal sqrt 
+        m = 1 / torch.sqrt(x)
+        return m 
+
+
+circuit = MyModel()
+
+x = torch.empty(1, 8).uniform_(0, 1)
+
+out = circuit(x)
+
+print(out)
+
+torch.onnx.export(circuit, x, "network.onnx",
+                  export_params=True,        # store the trained parameter weights inside the model file
+                  opset_version=17,          # the ONNX version to export the model to
+                  do_constant_folding=True,  # whether to execute constant folding for optimization
+                  input_names=['input'],   # the model's input names
+                  output_names=['output'],  # the model's output names
+                  dynamic_axes={'input': {0: 'batch_size'},  # variable length axes
+                                'output': {0: 'batch_size'}})
+
+
+d1 = ((x).detach().numpy()).reshape([-1]).tolist()
+
+data = dict(
+    input_data=[d1],
+)
+
+# Serialize data into file:
+json.dump(data, open("input.json", 'w'))

examples/onnx/rsqrt/input.json

@@ -0,0 +1 @@
+{"input_data": [[0.8590779900550842, 0.4029041528701782, 0.6507361531257629, 0.9782488942146301, 0.37392884492874146, 0.6867020726203918, 0.11407750844955444, 0.362740159034729]]}

examples/onnx/rsqrt/network.onnx

@@ -0,0 +1,17 @@
+pytorch2.2.2:Ź

+$
+input/Sqrt_output_0/Sqrt"Sqrt
+1
+/Sqrt_output_0output/Reciprocal"
+Reciprocal
+main_graphZ!
+input
+

+
+batch_size
+b"
+output
+

+
+batch_size
+B

src/circuit/ops/hybrid.rs

@@ -16,7 +16,6 @@ pub enum HybridOp {
     Recip {
         input_scale: utils::F32,
         output_scale: utils::F32,
-        use_range_check_for_int: bool,
     },
     Div {
         denom: utils::F32,
@@ -102,10 +101,9 @@ impl<F: PrimeField + TensorType + PartialOrd + std::hash::Hash> Op<F> for Hybrid
             HybridOp::Recip {
                 input_scale,
                 output_scale,
-                use_range_check_for_int,
             } => format!(
-                "RECIP (input_scale={}, output_scale={}, use_range_check_for_int={})",
-                input_scale, output_scale, use_range_check_for_int
+                "RECIP (input_scale={}, output_scale={})",
+                input_scale, output_scale
             ),
             HybridOp::Div {
                 denom,
@@ -187,31 +185,13 @@ impl<F: PrimeField + TensorType + PartialOrd + std::hash::Hash> Op<F> for Hybrid
             HybridOp::Recip {
                 input_scale,
                 output_scale,
-                use_range_check_for_int,
-            } => {
-                if input_scale.0.fract() == 0.0
-                    && output_scale.0.fract() == 0.0
-                    && *use_range_check_for_int
-                {
-                    layouts::recip(
-                        config,
-                        region,
-                        values[..].try_into()?,
-                        integer_rep_to_felt(input_scale.0 as i128),
-                        integer_rep_to_felt(output_scale.0 as i128),
-                    )?
-                } else {
-                    layouts::nonlinearity(
-                        config,
-                        region,
-                        values.try_into()?,
-                        &LookupOp::Recip {
-                            input_scale: *input_scale,
-                            output_scale: *output_scale,
-                        },
-                    )?
-                }
-            }
+            } => layouts::recip(
+                config,
+                region,
+                values[..].try_into()?,
+                integer_rep_to_felt(input_scale.0 as i128),
+                integer_rep_to_felt(output_scale.0 as i128),
+            )?,
             HybridOp::Div {
                 denom,
                 use_range_check_for_int,

src/circuit/ops/lookup.rs

@@ -15,89 +15,32 @@ use halo2curves::ff::PrimeField;
 /// An enum representing the operations that can be used to express more complex operations via accumulation
 #[derive(Clone, Debug, PartialEq, Eq, Hash, PartialOrd, Ord, Deserialize, Serialize)]
 pub enum LookupOp {
-    Div {
-        denom: utils::F32,
-    },
-    Cast {
-        scale: utils::F32,
-    },
-    Ceil {
-        scale: utils::F32,
-    },
-    Floor {
-        scale: utils::F32,
-    },
-    Round {
-        scale: utils::F32,
-    },
-    RoundHalfToEven {
-        scale: utils::F32,
-    },
-    Sqrt {
-        scale: utils::F32,
-    },
-    Rsqrt {
-        scale: utils::F32,
-    },
-    Recip {
-        input_scale: utils::F32,
-        output_scale: utils::F32,
-    },
-    Sigmoid {
-        scale: utils::F32,
-    },
-    Ln {
-        scale: utils::F32,
-    },
-    Exp {
-        scale: utils::F32,
-    },
-    Cos {
-        scale: utils::F32,
-    },
-    ACos {
-        scale: utils::F32,
-    },
-    Cosh {
-        scale: utils::F32,
-    },
-    ACosh {
-        scale: utils::F32,
-    },
-    Sin {
-        scale: utils::F32,
-    },
-    ASin {
-        scale: utils::F32,
-    },
-    Sinh {
-        scale: utils::F32,
-    },
-    ASinh {
-        scale: utils::F32,
-    },
-    Tan {
-        scale: utils::F32,
-    },
-    ATan {
-        scale: utils::F32,
-    },
-    Tanh {
-        scale: utils::F32,
-    },
-    ATanh {
-        scale: utils::F32,
-    },
-    Erf {
-        scale: utils::F32,
-    },
-    Pow {
-        scale: utils::F32,
-        a: utils::F32,
-    },
-    HardSwish {
-        scale: utils::F32,
-    },
+    Div { denom: utils::F32 },
+    Cast { scale: utils::F32 },
+    Ceil { scale: utils::F32 },
+    Floor { scale: utils::F32 },
+    Round { scale: utils::F32 },
+    RoundHalfToEven { scale: utils::F32 },
+    Sqrt { scale: utils::F32 },
+    Rsqrt { scale: utils::F32 },
+    Sigmoid { scale: utils::F32 },
+    Ln { scale: utils::F32 },
+    Exp { scale: utils::F32 },
+    Cos { scale: utils::F32 },
+    ACos { scale: utils::F32 },
+    Cosh { scale: utils::F32 },
+    ACosh { scale: utils::F32 },
+    Sin { scale: utils::F32 },
+    ASin { scale: utils::F32 },
+    Sinh { scale: utils::F32 },
+    ASinh { scale: utils::F32 },
+    Tan { scale: utils::F32 },
+    ATan { scale: utils::F32 },
+    Tanh { scale: utils::F32 },
+    ATanh { scale: utils::F32 },
+    Erf { scale: utils::F32 },
+    Pow { scale: utils::F32, a: utils::F32 },
+    HardSwish { scale: utils::F32 },
 }
 
 impl LookupOp {
@@ -118,10 +61,6 @@ impl LookupOp {
             LookupOp::Pow { scale, a } => format!("pow_{}_{}", scale, a),
             LookupOp::Div { denom } => format!("div_{}", denom),
             LookupOp::Cast { scale } => format!("cast_{}", scale),
-            LookupOp::Recip {
-                input_scale,
-                output_scale,
-            } => format!("recip_{}_{}", input_scale, output_scale),
             LookupOp::Sigmoid { scale } => format!("sigmoid_{}", scale),
             LookupOp::Sqrt { scale } => format!("sqrt_{}", scale),
             LookupOp::Rsqrt { scale } => format!("rsqrt_{}", scale),
@@ -173,14 +112,6 @@ impl LookupOp {
                 LookupOp::Cast { scale } => Ok::<_, TensorError>(
                     tensor::ops::nonlinearities::const_div(&x, f32::from(*scale).into()),
                 ),
-                LookupOp::Recip {
-                    input_scale,
-                    output_scale,
-                } => Ok::<_, TensorError>(tensor::ops::nonlinearities::recip(
-                    &x,
-                    input_scale.into(),
-                    output_scale.into(),
-                )),
                 LookupOp::Sigmoid { scale } => {
                     Ok::<_, TensorError>(tensor::ops::nonlinearities::sigmoid(&x, scale.into()))
                 }
@@ -260,13 +191,6 @@ impl<F: PrimeField + TensorType + PartialOrd + std::hash::Hash> Op<F> for Lookup
             LookupOp::Round { scale } => format!("ROUND(scale={})", scale),
             LookupOp::RoundHalfToEven { scale } => format!("ROUND_HALF_TO_EVEN(scale={})", scale),
             LookupOp::Pow { a, scale } => format!("POW(scale={}, exponent={})", scale, a),
-            LookupOp::Recip {
-                input_scale,
-                output_scale,
-            } => format!(
-                "RECIP(input_scale={}, output_scale={})",
-                input_scale, output_scale
-            ),
             LookupOp::Div { denom, .. } => format!("DIV(denom={})", denom),
             LookupOp::Cast { scale } => format!("CAST(scale={})", scale),
             LookupOp::Ln { scale } => format!("LN(scale={})", scale),
@@ -312,7 +236,6 @@ impl<F: PrimeField + TensorType + PartialOrd + std::hash::Hash> Op<F> for Lookup
                 let in_scale = inputs_scale[0];
                 in_scale + multiplier_to_scale(1. / scale.0 as f64)
             }
-            LookupOp::Recip { output_scale, .. } => multiplier_to_scale(output_scale.into()),
             _ => inputs_scale[0],
         };
         Ok(scale)

src/graph/utilities.rs

@@ -809,7 +809,6 @@ pub fn new_op_from_onnx(
             SupportedOp::Hybrid(HybridOp::Recip {
                 input_scale: (scale_to_multiplier(in_scale) as f32).into(),
                 output_scale: (scale_to_multiplier(max_scale) as f32).into(),
-                use_range_check_for_int: true,
             })
         }
 
@@ -1107,10 +1106,17 @@ pub fn new_op_from_onnx(
                 if c.raw_values.len() > 1 {
                     unimplemented!("only support scalar pow")
                 }
-                SupportedOp::Nonlinear(LookupOp::Pow {
-                    scale: scale_to_multiplier(inputs[0].out_scales()[0]).into(),
-                    a: crate::circuit::utils::F32(c.raw_values[0]),
-                })
+
+                let exponent = c.raw_values[0];
+
+                if exponent.fract() == 0.0 {
+                    SupportedOp::Linear(PolyOp::Pow(exponent as u32))
+                } else {
+                    SupportedOp::Nonlinear(LookupOp::Pow {
+                        scale: scale_to_multiplier(inputs[0].out_scales()[0]).into(),
+                        a: crate::circuit::utils::F32(exponent),
+                    })
+                }
             } else {
                 unimplemented!("only support constant pow for now")
             }

tests/integration_tests.rs

@@ -205,7 +205,7 @@ mod native_tests {
         "1l_tiny_div",
     ];
 
-    const TESTS: [&str; 94] = [
+    const TESTS: [&str; 95] = [
         "1l_mlp", //0
         "1l_slice",
         "1l_concat",
@@ -304,6 +304,7 @@ mod native_tests {
         "lstm_large",  // 91
         "lstm_medium", // 92
         "lenet_5",     // 93
+        "rsqrt",       // 94
     ];
 
     const WASM_TESTS: [&str; 46] = [
@@ -542,7 +543,7 @@ mod native_tests {
             }
         });
 
-            seq!(N in 0..=93 {
+            seq!(N in 0..=94 {
 
             #(#[test_case(TESTS[N])])*
             #[ignore]
@@ -1118,7 +1119,7 @@ mod native_tests {
 
             });
 
-            seq!(N in 0..=93 {
+            seq!(N in 0..4 {
                 #(#[test_case(TESTS[N])])*
                 fn kzg_evm_prove_and_verify_reusable_verifier_(test: &str) {
                     crate::native_tests::init_binary();