fix: Fix nondeterministic output tensor selection

.github/workflows/test.yml

@@ -5,7 +5,7 @@ on:
 env:
   CARGO_TERM_COLOR: always
   RUSTFLAGS: "-Dwarnings"
-  ONNX_VERSION: v1.22.0
+  ONNX_VERSION: v1.22.1
 
 jobs:
   test:

src/output/embedding_output.rs

@@ -10,7 +10,7 @@ use super::{OutputKey, OutputPrecedence};
 /// pooling etc. This struct should contain all the necessary information for the
 /// post-processing to be performed.
 pub struct SingleBatchOutput {
-    pub outputs: std::collections::HashMap<String, ort::value::Value>,
+    pub outputs: std::collections::BTreeMap<String, ort::value::Value>,
     pub attention_mask_array: Array2<i64>,
 }
 
@@ -26,7 +26,14 @@ impl SingleBatchOutput {
         let ort_output: &ort::value::Value = precedence
             .key_precedence()
             .find_map(|key| match key {
-                OutputKey::OnlyOne => self.outputs.values().next(),
+                // Only select the sole output if and only if there is exactly one.
+                OutputKey::OnlyOne => {
+                    if self.outputs.len() == 1 {
+                        self.outputs.values().next()
+                    } else {
+                        None
+                    }
+                }
                 OutputKey::ByOrder(idx) => self.outputs.values().nth(*idx),
                 OutputKey::ByName(name) => self.outputs.get(*name),
             })

tests/embeddings.rs

@@ -659,3 +659,24 @@ fn test_allminilml6v2_match_python_counterpart() {
         assert!((expected - actual).abs() < tolerance);
     }
 }
+
+// Ref: https://github.com/Anush008/fastembed-rs/issues/171#issue-3209484009
+#[test]
+fn clip_vit_b32_deterministic_across_calls() {
+    let q = "red car";
+    let mut fe = TextEmbedding::try_new(InitOptions::new(EmbeddingModel::ClipVitB32)).unwrap();
+    let mut first: Option<Vec<f32>> = None;
+    for i in 0..100 {
+        let vecs = fe.embed(vec![q], None).unwrap();
+        if first.is_none() {
+            first = Some(vecs[0].clone());
+        } else {
+            assert_eq!(
+                vecs[0],
+                *first.as_ref().unwrap(),
+                "Embedding changed after {} iterations",
+                i
+            );
+        }
+    }
+}