diff --git a/test-rt/infra/src/lib.rs b/test-rt/infra/src/lib.rs
index e9e2213c82..c5aaebc95a 100644
--- a/test-rt/infra/src/lib.rs
+++ b/test-rt/infra/src/lib.rs
@@ -1,11 +1,10 @@
 #![allow(clippy::len_zero)]
 use std::collections::HashMap;
 use std::io::Write;
-use std::marker::PhantomData;
 
 use dyn_clone::DynClone;
 use itertools::Itertools;
-use proptest::prelude::{any, any_with, Arbitrary};
+use proptest::prelude::{any_with, Arbitrary};
 use proptest::test_runner::{Config, FileFailurePersistence, TestRunner};
 use tract_core::runtime::Runtime;
 use tract_core::tract_data::TractResult;
@@ -51,8 +50,11 @@ impl TestSuite {
         }
     }
 
-    pub fn add_arbitrary<A: Arbitrary + Test + Clone>(&mut self, id: impl ToString, params: A::Parameters)
-    where
+    pub fn add_arbitrary<A: Arbitrary + Test + Clone>(
+        &mut self,
+        id: impl ToString,
+        params: A::Parameters,
+    ) where
         A::Parameters: Clone + Send + Sync,
     {
         self.add(id, ProptestWrapper::<A>(params));
diff --git a/test-rt/suite-conv/src/conv_f32.rs b/test-rt/suite-conv/src/conv_f32.rs
index 2a463865a2..e2c558ff94 100644
--- a/test-rt/suite-conv/src/conv_f32.rs
+++ b/test-rt/suite-conv/src/conv_f32.rs
@@ -1,9 +1,17 @@
+use std::ops::Range;
+
 use super::*;
 use infra::*;
 use proptest::collection::vec;
-use proptest::test_runner::{Config, FileFailurePersistence, TestRunner};
 use tract_itertools::izip;
 
+#[derive(Debug, Clone, Default)]
+pub struct ConvProblemParams {
+    pub no_group: bool,
+    pub no_arbitrary_grouping: bool,
+    pub geo_rank: Option<Range<usize>>,
+}
+
 #[derive(Debug, Clone)]
 pub struct ConvProblem {
     pub shape_in: DataShape,
@@ -30,21 +38,41 @@ impl ConvProblem {
             KernelFormat::HWIO => self.kernel.shape()[self.kernel.ndim() - 1],
             KernelFormat::OHWI => self.kernel.shape()[0],
         };
-        let (shape_out, left_pads): (TVec<_>, TVec<_>) = if self.pad == PaddingSpec::Valid {
-            izip!(self.shape_in.hw_dims(), self.geo_ker(), &self.strides)
+        let (shape_out, left_pads): (TVec<_>, TVec<_>) = match &self.pad {
+            PaddingSpec::Valid => izip!(self.shape_in.hw_dims(), self.geo_ker(), &self.strides)
                 .map(|(i, k, s)| {
                     let out = (*i + 1).saturating_sub(*k).divceil(*s);
                     (out, 0)
                 })
-                .unzip()
-        } else {
-            izip!(self.shape_in.hw_dims(), self.geo_ker(), &self.strides)
+                .unzip(),
+            PaddingSpec::SameUpper => izip!(self.shape_in.hw_dims(), self.geo_ker(), &self.strides)
                 .map(|(input, k, stride)| {
                     let out = input.divceil(*stride);
                     let pad = ((out - 1) * stride + k).saturating_sub(*input);
                     (out, pad / 2)
                 })
-                .unzip()
+                .unzip(),
+            PaddingSpec::SameLower => izip!(self.shape_in.hw_dims(), self.geo_ker(), &self.strides)
+                .map(|(input, k, stride)| {
+                    let out = input.divceil(*stride);
+                    let pad = ((out - 1) * stride + k).saturating_sub(*input);
+                    (out, pad.divceil(2))
+                })
+                .unzip(),
+            PaddingSpec::Explicit(l, r, ceil) => {
+                izip!(self.shape_in.hw_dims(), self.geo_ker(), &self.strides, l, r)
+                    .map(|(input, k, stride, l, r)| {
+                        let dil = 1;
+                        let kf = (k - 1) * dil + 1;
+                        let out = if *ceil {
+                            (input + l + r).saturating_sub(kf).divceil(*stride) + 1
+                        } else {
+                            (input + l + r).saturating_sub(kf) / *stride + 1
+                        };
+                        (out, *l)
+                    })
+                    .unzip()
+            }
         };
         let shape_out = self
             .shape_in
@@ -144,15 +172,11 @@ impl ConvProblem {
     }
 }
 
-#[derive(Debug, Clone, Default)]
-pub struct ConvProblemParams {
-    pub no_arbitrary_grouping: bool,
-}
-
 impl Arbitrary for ConvProblem {
     type Parameters = ConvProblemParams;
     type Strategy = BoxedStrategy<ConvProblem>;
     fn arbitrary_with(params: Self::Parameters) -> Self::Strategy {
+        let geo_rank = params.geo_rank.unwrap_or(1..4);
         (
             data_format(),
             kernel_format(),
@@ -160,8 +184,8 @@ impl Arbitrary for ConvProblem {
             1usize..=3,
             1usize..=4,
             1usize..=4,
-            1usize..=3,
-            (1usize..=3).prop_flat_map(shapes),
+            1usize..=(if params.no_group { 1 } else { 3 }),
+            geo_rank.prop_flat_map(shapes),
         )
             .prop_flat_map(
                 move |(df, kf, pad, n, mut ci0, mut co0, group, (mut ker_shape, data_shape))| {
@@ -217,28 +241,15 @@ impl Test for ConvProblem {
         let mut output =
             runtime.prepare(self.tract()?)?.run(tvec![self.data.clone().into_tvalue()])?;
         let output = output.remove(0).into_tensor();
-        reference.close_enough(&output, true)
-    }
-}
-
-#[derive(Clone)]
-pub struct ConvProptest;
-
-impl Test for ConvProptest {
-    fn run(&self, runtime: &dyn Runtime) -> TestResult {
-        let mut runner = TestRunner::new(Config {
-            failure_persistence: Some(Box::new(FileFailurePersistence::Off)),
-            ..Config::default()
-        });
-        runner.run(&any::<ConvProblem>(), |v| Ok(v.run(runtime).unwrap()))?;
-        Ok(())
+        eprintln!("output: {output:?} reference: {reference:?}");
+        output.close_enough(&reference, true)
     }
 }
 
 pub fn suite() -> TractResult<TestSuite> {
     let mut suite = TestSuite::default();
 
-    suite.add("proptest", ConvProptest);
+    suite.add_arbitrary::<ConvProblem>("proptest", ConvProblemParams::default());
 
     suite.add(
         "trivial_0",
@@ -308,6 +319,21 @@ pub fn suite() -> TractResult<TestSuite> {
             strides: tvec!(1),
         },
     );
+
+    suite.add(
+        "group_0",
+        ConvProblem {
+            shape_in: DataFormat::CHW.from_n_c_hw(1, 2, [1])?,
+            kernel_format: KernelFormat::OIHW,
+            group: 2,
+            data: arr2(&[[0.0f32], [0.0]]).into_dyn(),
+            kernel: arr3(&[[[0.0f32]], [[0.0]]]).into_dyn(),
+            bias: None,
+            pad: PaddingSpec::Valid,
+            strides: tvec!(1),
+        },
+    );
+
     suite.add(
         "group_1",
         ConvProblem {
@@ -627,7 +653,7 @@ pub fn suite() -> TractResult<TestSuite> {
     );
 
     suite.add(
-        "same_0",
+        "same_1d_0",
         ConvProblem {
             shape_in: DataFormat::HWC.from_n_c_hw(1, 1, [1])?,
             kernel_format: KernelFormat::OIHW,
@@ -641,13 +667,13 @@ pub fn suite() -> TractResult<TestSuite> {
     );
 
     suite.add(
-        "same_1",
+        "same_1d_1",
         ConvProblem {
             shape_in: DataFormat::HWC.from_n_c_hw(1, 1, [2])?,
             kernel_format: KernelFormat::OIHW,
             group: 1,
             data: arr2(&[[0.0], [1.0]]).into_dyn(),
-            kernel: arr3(&[[[0.0, 1.0]]]).into_dyn(),
+            kernel: arr3(&[[[0.0, 2.0]]]).into_dyn(),
             bias: None,
             pad: PaddingSpec::SameUpper,
             strides: tvec!(1),
diff --git a/test-rt/suite-conv/tests/proptest.proptest-regressions b/test-rt/suite-conv/tests/proptest.proptest-regressions
deleted file mode 100644
index c46a644fd1..0000000000
--- a/test-rt/suite-conv/tests/proptest.proptest-regressions
+++ /dev/null
@@ -1,7 +0,0 @@
-# Seeds for failure cases proptest has generated in the past. It is
-# automatically read and these particular cases re-run before any
-# novel cases are generated.
-#
-# It is recommended to check this file in to source control so that
-# everyone who runs the test benefits from these saved cases.
-cc fbf47a092afc0fff944e6c5a5d7fb4f401641b72bf0913ca7cdbafb3460c123f # shrinks to pb = ConvProblem { shape_in: HWC 5,7,8 (strides: 56,8,1), kernel_format: OHWI, group: 2, data: [[[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]],   [[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]],   [[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]],   [[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]],   [[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0],   [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]]], shape=[5, 7, 8], strides=[56, 8, 1], layout=Cc (0x5), dynamic ndim=3, kernel: [[[[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]],    [[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]]],    [[[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]],    [[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],    [0.0, 0.0, 0.0, -2.0, 8.0, 8.0, -9.0, -2.0],    [6.0, 8.0, -9.0, -9.0, 10.0, 10.0, -10.0, 6.0]]],    [[[-3.0, 9.0, -5.0, -3.0, -6.0, -1.0, 7.0, -9.0],    [10.0, -6.0, -6.0, 9.0, 0.0, -2.0, -2.0, 4.0],    [-5.0, -5.0, 4.0, -5.0, -9.0, -7.0, -1.0, -4.0]],    [[10.0, -4.0, -5.0, -10.0, 0.0, 6.0, -4.0, -2.0],    [6.0, -2.0, -9.0, 0.0, -7.0, 8.0, -10.0, 0.0],    [7.0, -8.0, -7.0, -8.0, -6.0, 3.0, 0.0, 10.0]]]], shape=[3, 2, 3, 8], strides=[48, 24, 8, 1], layout=Cc (0x5), dynamic ndim=4, bias: None, pad: Valid, strides: [1, 3] }
diff --git a/test-rt/test-onnx-nnef-cycle/src/lib.rs b/test-rt/test-onnx-nnef-cycle/src/lib.rs
index e519a8a05e..50dc8bf143 100644
--- a/test-rt/test-onnx-nnef-cycle/src/lib.rs
+++ b/test-rt/test-onnx-nnef-cycle/src/lib.rs
@@ -16,7 +16,7 @@ impl Runtime for NnefCyclingRuntime {
         self.0.write_to_tar(&model, &mut buffer)?;
         info!("Reload from NNEF");
         let reloaded = self.0.model_for_read(&mut &*buffer)?;
-        Ok(Box::new(reloaded.into_optimized()?.into_runnable()?))
+        Ok(Box::new(Arc::new(reloaded.into_optimized()?.into_runnable()?)))
     }
 }
 
diff --git a/test-rt/test-tflite/src/tflite_runtime.rs b/test-rt/test-tflite/src/tflite_runtime.rs
index d8a450836f..33a8a438fa 100644
--- a/test-rt/test-tflite/src/tflite_runtime.rs
+++ b/test-rt/test-tflite/src/tflite_runtime.rs
@@ -14,6 +14,7 @@ impl Runtime for TfliteRuntime {
     fn prepare(&self, model: TypedModel) -> TractResult<Box<dyn Runnable>> {
         let mut buffer = vec![];
         self.0.write(&model, &mut buffer)?;
+ //       std::fs::write("foo.tflite", &buffer).unwrap();
         Ok(Box::new(TfliteRunnable(buffer)))
     }
 }
@@ -39,7 +40,7 @@ impl State for TfliteState {
         for (ix, input) in inputs.iter().enumerate() {
             let input_ix = self.0.inputs()[ix];
             let input_tensor = self.0.tensor_info(input_ix).unwrap();
-            assert_eq!(input_tensor.element_kind as u32, 1);
+            assert_eq!(input_tensor.element_kind as u32, 1); // 1 is f32
             assert_eq!(input_tensor.dims, input.shape());
             self.0.tensor_buffer_mut(input_ix).unwrap().copy_from_slice(unsafe { input.as_bytes() })
         }
diff --git a/test-rt/test-tflite/suite.rs b/test-rt/test-tflite/suite.rs
index 61d6ed6403..6cf1b2fc90 100644
--- a/test-rt/test-tflite/suite.rs
+++ b/test-rt/test-tflite/suite.rs
@@ -7,7 +7,7 @@ pub fn suite() -> infra::TestSuite {
     conv.ignore(&ignore_conv);
     conv.add_arbitrary::<ConvProblem>(
         "proptest",
-        ConvProblemParams { no_arbitrary_grouping: true, ..ConvProblemParams::default() },
+        ConvProblemParams { no_group: true, geo_rank: Some(1..3), ..ConvProblemParams::default() },
     );
     infra::TestSuite::default().with("onnx", onnx).with("conv", conv)
 }
@@ -29,4 +29,6 @@ fn ignore_conv(t: &[String]) -> bool {
         || unit == "lazy_im2col_big_2"
         || unit == "batch_3d"
         || unit == "bias_3d_1"
+        // nonsense. bug in tfl ? hole in the spec ?
+        || unit == "same_1d_1"
 }
diff --git a/tflite/src/rewriter.rs b/tflite/src/rewriter.rs
index 69c696b1d3..35c34de537 100644
--- a/tflite/src/rewriter.rs
+++ b/tflite/src/rewriter.rs
@@ -64,13 +64,13 @@ fn make_1d_2d(
     conv: &ConvUnary,
 ) -> TractResult<Option<TypedModelPatch>> {
     if conv.pool_spec.rank() == 1 {
-        let pos = conv.pool_spec.data_format.h_axis();
+        let pos = conv.pool_spec.data_format.h_axis() + 1;
         let mut new = conv.clone();
-        new.pool_spec.kernel_shape.insert(0, 1);
-        new.pool_spec.dilations.iter_mut().for_each(|dil| dil.insert(0, 1));
-        new.pool_spec.strides.iter_mut().for_each(|dil| dil.insert(0, 1));
+        new.pool_spec.kernel_shape.insert(1, 1);
+        new.pool_spec.dilations.iter_mut().for_each(|dil| dil.insert(1, 1));
+        new.pool_spec.strides.iter_mut().for_each(|dil| dil.insert(1, 1));
         let mut kernel = new.kernel.clone().into_tensor();
-        kernel.insert_axis(conv.kernel_fmt.h_axis())?;
+        kernel.insert_axis(conv.kernel_fmt.h_axis() + 1)?;
         new.kernel = kernel.into_arc_tensor();
         let mut patch = TypedModelPatch::default();
         let mut wire = tvec!(patch.tap_model(model, node.inputs[0])?);