apr-cli 0.4.16

CLI tool for APR model inspection, debugging, and operations
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221


    #[test]
    fn test_print_tensor_diff_row_medium_diff() {
        let stats = TensorValueStats {
            name: "ffn.weight".to_string(),
            shape_a: vec![32, 32],
            shape_b: vec![32, 32],
            element_count: 1024,
            mean_diff: 0.03,
            max_diff: 0.08,
            rmse: 0.04,
            cosine_similarity: 0.998,
            identical_count: 100,
            small_diff_count: 200,
            medium_diff_count: 500,
            large_diff_count: 224,
            status: TensorDiffStatus::MediumDiff,
        };
        print_tensor_diff_row(&stats);
    }

    #[test]
    fn test_print_tensor_diff_row_large_diff() {
        let stats = TensorValueStats {
            name: "lm_head.weight".to_string(),
            shape_a: vec![50, 50],
            shape_b: vec![50, 50],
            element_count: 2500,
            mean_diff: 0.3,
            max_diff: 0.9,
            rmse: 0.4,
            cosine_similarity: 0.95,
            identical_count: 0,
            small_diff_count: 0,
            medium_diff_count: 500,
            large_diff_count: 2000,
            status: TensorDiffStatus::LargeDiff,
        };
        print_tensor_diff_row(&stats);
    }

    #[test]
    fn test_print_tensor_diff_row_small_diff() {
        let stats = TensorValueStats {
            name: "embed.weight".to_string(),
            shape_a: vec![16, 16],
            shape_b: vec![16, 16],
            element_count: 256,
            mean_diff: 0.003,
            max_diff: 0.008,
            rmse: 0.004,
            cosine_similarity: 0.9999,
            identical_count: 50,
            small_diff_count: 150,
            medium_diff_count: 56,
            large_diff_count: 0,
            status: TensorDiffStatus::SmallDiff,
        };
        print_tensor_diff_row(&stats);
    }

    #[test]
    fn test_print_tensor_diff_row_long_name_truncation() {
        let long_name = "model.layers.99.self_attn.q_proj.weight.extra.suffix.that.is.very.long";
        let stats = TensorValueStats {
            name: long_name.to_string(),
            shape_a: vec![4],
            shape_b: vec![4],
            element_count: 4,
            mean_diff: 0.0,
            max_diff: 0.0,
            rmse: 0.0,
            cosine_similarity: 1.0,
            identical_count: 4,
            small_diff_count: 0,
            medium_diff_count: 0,
            large_diff_count: 0,
            status: TensorDiffStatus::Identical,
        };
        // Exercises truncate_str for name > 40 chars
        print_tensor_diff_row(&stats);
    }

    #[test]
    fn test_print_tensor_diff_row_cosine_similarity_ranges() {
        // Test cosine similarity coloring: > 0.9999
        let make_stats = |cos: f32| TensorValueStats {
            name: "t".to_string(),
            shape_a: vec![4],
            shape_b: vec![4],
            element_count: 4,
            mean_diff: 0.01,
            max_diff: 0.02,
            rmse: 0.01,
            cosine_similarity: cos,
            identical_count: 0,
            small_diff_count: 4,
            medium_diff_count: 0,
            large_diff_count: 0,
            status: TensorDiffStatus::SmallDiff,
        };
        // > 0.9999 (green)
        print_tensor_diff_row(&make_stats(0.99999));
        // > 0.999 (blue)
        print_tensor_diff_row(&make_stats(0.9995));
        // > 0.99 (yellow)
        print_tensor_diff_row(&make_stats(0.995));
        // <= 0.99 (red)
        print_tensor_diff_row(&make_stats(0.5));
    }

    // ==================== DiffEntryJson struct ====================

    #[test]
    fn test_diff_entry_json_serialization() {
        let entry = DiffEntryJson {
            field: "tensor_count".to_string(),
            file1_value: "100".to_string(),
            file2_value: "200".to_string(),
            category: "tensor".to_string(),
            diff_type: "data".to_string(),
        };
        let json = serde_json::to_string(&entry).expect("serialize");
        assert!(json.contains("\"field\":\"tensor_count\""));
        assert!(json.contains("\"file1_value\":\"100\""));
        assert!(json.contains("\"file2_value\":\"200\""));
        assert!(json.contains("\"category\":\"tensor\""));
        assert!(json.contains("\"type\":\"data\""));
    }

    // ==================== DiffResultJson category mapping ====================

    #[test]
    fn test_diff_result_json_all_categories() {
        let report = DiffReport {
            path1: "a".to_string(),
            path2: "b".to_string(),
            format1: "APR".to_string(),
            format2: "GGUF".to_string(),
            differences: vec![
                DiffEntry {
                    field: "f1".to_string(),
                    value1: "v1".to_string(),
                    value2: "v2".to_string(),
                    category: DiffCategory::Format,
                },
                DiffEntry {
                    field: "f2".to_string(),
                    value1: "v1".to_string(),
                    value2: "v2".to_string(),
                    category: DiffCategory::Metadata,
                },
                DiffEntry {
                    field: "f3".to_string(),
                    value1: "v1".to_string(),
                    value2: "v2".to_string(),
                    category: DiffCategory::Tensor,
                },
                DiffEntry {
                    field: "f4".to_string(),
                    value1: "v1".to_string(),
                    value2: "v2".to_string(),
                    category: DiffCategory::Quantization,
                },
                DiffEntry {
                    field: "f5".to_string(),
                    value1: "v1".to_string(),
                    value2: "v2".to_string(),
                    category: DiffCategory::Size,
                },
            ],
            inspection1: None,
            inspection2: None,
        };
        let json = DiffResultJson::from(&report);
        assert_eq!(json.difference_count, 5);
        assert_eq!(json.differences[0].category, "format");
        assert_eq!(json.differences[1].category, "metadata");
        assert_eq!(json.differences[2].category, "tensor");
        assert_eq!(json.differences[3].category, "quantization");
        assert_eq!(json.differences[4].category, "size");
    }

    // ==================== Integration: compute_tensor_diff_stats + from_diff_info ====================

    #[test]
    fn test_compute_stats_transposed_identical_values() {
        // Simulate a tensor that's transposed but values happen to match linearly
        let data = vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0];
        let stats = compute_tensor_diff_stats("t", &[2, 3], &[3, 2], &data, &data, false);
        // All 6 elements identical linearly => ident_ratio = 1.0 > 0.99
        assert_eq!(stats.status, TensorDiffStatus::Transposed);
        assert_eq!(stats.identical_count, 6);
    }

    #[test]
    fn test_compute_stats_small_diff_boundary() {
        // max_diff exactly at SmallDiff boundary
        let data_a = vec![1.0, 2.0];
        let data_b = vec![1.001, 2.0]; // max_diff = 0.001 exactly
        let stats = compute_tensor_diff_stats("boundary", &[2], &[2], &data_a, &data_b, false);
        // 0.001 is NOT < 0.001, so SmallDiff
        assert_eq!(stats.status, TensorDiffStatus::SmallDiff);
    }

    #[test]
    fn test_compute_stats_one_element() {
        let stats = compute_tensor_diff_stats("single", &[1], &[1], &[42.0], &[42.0], false);
        assert_eq!(stats.status, TensorDiffStatus::Identical);
        assert_eq!(stats.element_count, 1);
        assert_eq!(stats.identical_count, 1);
        assert!((stats.cosine_similarity - 1.0).abs() < 1e-5);
    }

    #[test]
    fn test_compute_stats_one_element_a_only() {
        // data_b is empty => element_count = min(1, 0) = 0 => early return Critical
        let stats = compute_tensor_diff_stats("asym", &[1], &[0], &[1.0], &[], false);
        assert_eq!(stats.status, TensorDiffStatus::Critical);
        assert_eq!(stats.element_count, 0);
    }