Error on empty set, backward compatibility (13 and below) with 'axes'

huggingface · Oct 15, 2024 · b40491a · b40491a
1 parent 4f8cd5c
commit b40491a
Show file tree

Hide file tree

Showing 2 changed files with 174 additions and 20 deletions.
diff --git a/candle-onnx/src/eval.rs b/candle-onnx/src/eval.rs
@@ -1190,17 +1190,26 @@ fn simple_eval_(
                 values.insert(node.output[0].clone(), out);
             }
             // https://onnx.ai/onnx/operators/onnx__ReduceMax.html#reducemax
-            // Version 18 impl
             "ReduceMax" => {
                 let input = get(&node.input[0])?;
                 let axes = get_opt(1);
                 let keepdims = get_attr_opt::<i64>(node, "keepdims")?.copied().unwrap_or(1) == 1;
 
                 let axes = if let Some(Ok(axes)) = axes {
+                    // Satisfies version 18+
+                    axes.to_vec1::<i64>().ok()
+                } else if let Ok(Some(axes)) = get_attr_opt::<[i64]>(node, "axes") {
+                    // Backward compatiblity with version 13 and below
+                    Some(axes.to_vec())
+                } else {
+                    None
+                };
+
+                let axes = if let Some(axes) = axes {
                     let rank = input.rank();
                     let mut axes_set = HashSet::new();
+
                     let mut axes = axes
-                        .to_vec1::<i64>()?
                         .iter()
                         .map(|a| {
                             let axis = if *a < 0 {
@@ -1230,7 +1239,10 @@ fn simple_eval_(
                 // TODO: Handle empty set
                 // Definition:
                 // "Reduction over an empty set of values yields minus infinity (if supported by the datatype) or the minimum value of the data type otherwise"
-                // Numpy yields error: ValueError: zero-size array to reduction operation maximum which has no identity
+                // For now, this will throw an error
+                if input.elem_count() == 0 {
+                    bail!("reduction over zero-size tensor not supported");
+                }
 
                 let output = if let Some(axes) = axes {
                     let mut result = input.clone();
@@ -1286,18 +1298,27 @@ fn simple_eval_(
                 };
                 values.insert(node.output[0].clone(), output);
             }
-            // https://onnx.ai/onnx/operators/onnx__ReduceMin.html#reducemin-20
-            // Version 18 impl
+            // https://onnx.ai/onnx/operators/onnx__ReduceMin.html#reducemin
             "ReduceMin" => {
                 let input = get(&node.input[0])?;
                 let axes = get_opt(1);
                 let keepdims = get_attr_opt::<i64>(node, "keepdims")?.copied().unwrap_or(1) == 1;
 
                 let axes = if let Some(Ok(axes)) = axes {
+                    // Satisfies version 18+
+                    axes.to_vec1::<i64>().ok()
+                } else if let Ok(Some(axes)) = get_attr_opt::<[i64]>(node, "axes") {
+                    // Backward compatiblity with version 13 and below
+                    Some(axes.to_vec())
+                } else {
+                    None
+                };
+
+                let axes = if let Some(axes) = axes {
                     let rank = input.rank();
                     let mut axes_set = HashSet::new();
+
                     let mut axes = axes
-                        .to_vec1::<i64>()?
                         .iter()
                         .map(|a| {
                             let axis = if *a < 0 {
@@ -1326,8 +1347,11 @@ fn simple_eval_(
 
                 // TODO: Handle empty set
                 // Definition:
-                // "Reduction over an empty set of values yields minus infinity (if supported by the datatype) or the minimum value of the data type otherwise"
-                // Numpy yields error: ValueError: zero-size array to reduction operation maximum which has no identity
+                // "Reduction over an empty set of values yields positive infinity (if supported by the datatype) or the max value of the data type otherwise"
+                // For now, this will throw an error
+                if input.elem_count() == 0 {
+                    bail!("reduction over zero-size tensor not supported");
+                }
 
                 let output = if let Some(axes) = axes {
                     let mut result = input.clone();