hf-transformers-bot commited on
Commit
67035f9
·
verified ·
1 Parent(s): c46c446

Upload 2025-11-15/ci_results_run_models_gpu/new_failures.json with huggingface_hub

Browse files
2025-11-15/ci_results_run_models_gpu/new_failures.json ADDED
@@ -0,0 +1,1557 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "apertus": {
3
+ "single-gpu": [
4
+ "tests/models/apertus/test_modeling_apertus.py::ApertusModelTest::test_flash_attn_2_fp32_ln"
5
+ ],
6
+ "multi-gpu": [
7
+ "tests/models/apertus/test_modeling_apertus.py::ApertusModelTest::test_flash_attn_2_fp32_ln"
8
+ ]
9
+ },
10
+ "arcee": {
11
+ "single-gpu": [
12
+ "tests/models/arcee/test_modeling_arcee.py::ArceeModelTest::test_flash_attn_2_fp32_ln"
13
+ ],
14
+ "multi-gpu": [
15
+ "tests/models/arcee/test_modeling_arcee.py::ArceeModelTest::test_flash_attn_2_fp32_ln"
16
+ ]
17
+ },
18
+ "aria": {
19
+ "single-gpu": [
20
+ "tests/models/aria/test_modeling_aria.py::AriaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
21
+ ],
22
+ "multi-gpu": [
23
+ "tests/models/aria/test_modeling_aria.py::AriaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
24
+ ]
25
+ },
26
+ "audio_spectrogram_transformer": {
27
+ "single-gpu": [
28
+ "tests/models/audio_spectrogram_transformer/test_modeling_audio_spectrogram_transformer.py::ASTModelTest::test_flash_attn_2_can_dispatch_composite_models"
29
+ ],
30
+ "multi-gpu": [
31
+ "tests/models/audio_spectrogram_transformer/test_modeling_audio_spectrogram_transformer.py::ASTModelTest::test_flash_attn_2_can_dispatch_composite_models"
32
+ ]
33
+ },
34
+ "audioflamingo3": {
35
+ "single-gpu": [
36
+ "tests/models/audioflamingo3/test_modeling_audioflamingo3.py::AudioFlamingo3ForConditionalGenerationModelTest::test_eager_matches_fa2_generate",
37
+ "tests/models/audioflamingo3/test_modeling_audioflamingo3.py::AudioFlamingo3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
38
+ ],
39
+ "multi-gpu": [
40
+ "tests/models/audioflamingo3/test_modeling_audioflamingo3.py::AudioFlamingo3ForConditionalGenerationModelTest::test_eager_matches_fa2_generate",
41
+ "tests/models/audioflamingo3/test_modeling_audioflamingo3.py::AudioFlamingo3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
42
+ ]
43
+ },
44
+ "aya_vision": {
45
+ "single-gpu": [
46
+ "tests/models/aya_vision/test_modeling_aya_vision.py::AyaVisionModelTest::test_flash_attn_2_fp32_ln"
47
+ ],
48
+ "multi-gpu": [
49
+ "tests/models/aya_vision/test_modeling_aya_vision.py::AyaVisionModelTest::test_flash_attn_2_fp32_ln"
50
+ ]
51
+ },
52
+ "bamba": {
53
+ "single-gpu": [
54
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_fp32_ln",
55
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_inference_equivalence",
56
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_inference_equivalence_right_padding"
57
+ ],
58
+ "multi-gpu": [
59
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_fp32_ln",
60
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_inference_equivalence",
61
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_inference_equivalence_right_padding"
62
+ ]
63
+ },
64
+ "bark": {
65
+ "single-gpu": [
66
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_eager_matches_fa2_generate",
67
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_flash_attention_2_continue_generate_with_position_ids",
68
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_flash_attn_2_fp32_ln",
69
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_eager_matches_fa2_generate",
70
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_flash_attention_2_continue_generate_with_position_ids",
71
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_flash_attn_2_fp32_ln"
72
+ ],
73
+ "multi-gpu": [
74
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_eager_matches_fa2_generate",
75
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_flash_attention_2_continue_generate_with_position_ids",
76
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_flash_attn_2_fp32_ln",
77
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_eager_matches_fa2_generate",
78
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_flash_attention_2_continue_generate_with_position_ids",
79
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_flash_attn_2_fp32_ln"
80
+ ]
81
+ },
82
+ "bart": {
83
+ "single-gpu": [
84
+ "tests/models/bart/test_modeling_bart.py::BartModelTest::test_flash_attn_2_fp32_ln",
85
+ "tests/models/bart/test_modeling_bart.py::BartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
86
+ ],
87
+ "multi-gpu": [
88
+ "tests/models/bart/test_modeling_bart.py::BartModelTest::test_flash_attn_2_fp32_ln",
89
+ "tests/models/bart/test_modeling_bart.py::BartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
90
+ ]
91
+ },
92
+ "bert": {
93
+ "single-gpu": [
94
+ "tests/models/bert/test_modeling_bert.py::BertModelTest::test_flash_attn_2_fp32_ln"
95
+ ],
96
+ "multi-gpu": [
97
+ "tests/models/bert/test_modeling_bert.py::BertModelTest::test_flash_attn_2_fp32_ln"
98
+ ]
99
+ },
100
+ "bert_generation": {
101
+ "single-gpu": [
102
+ "tests/models/bert_generation/test_modeling_bert_generation.py::BertGenerationEncoderTest::test_flash_attn_2_fp32_ln"
103
+ ],
104
+ "multi-gpu": [
105
+ "tests/models/bert_generation/test_modeling_bert_generation.py::BertGenerationEncoderTest::test_flash_attn_2_fp32_ln"
106
+ ]
107
+ },
108
+ "biogpt": {
109
+ "single-gpu": [
110
+ "tests/models/biogpt/test_modeling_biogpt.py::BioGptModelTest::test_flash_attn_2_fp32_ln"
111
+ ],
112
+ "multi-gpu": [
113
+ "tests/models/biogpt/test_modeling_biogpt.py::BioGptModelTest::test_flash_attn_2_fp32_ln"
114
+ ]
115
+ },
116
+ "bitnet": {
117
+ "single-gpu": [
118
+ "tests/models/bitnet/test_modeling_bitnet.py::BitNetModelTest::test_flash_attn_2_fp32_ln"
119
+ ],
120
+ "multi-gpu": [
121
+ "tests/models/bitnet/test_modeling_bitnet.py::BitNetModelTest::test_flash_attn_2_fp32_ln"
122
+ ]
123
+ },
124
+ "blenderbot": {
125
+ "single-gpu": [
126
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotModelTest::test_flash_attn_2_fp32_ln",
127
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
128
+ ],
129
+ "multi-gpu": [
130
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotModelTest::test_flash_attn_2_fp32_ln",
131
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
132
+ ]
133
+ },
134
+ "blenderbot_small": {
135
+ "single-gpu": [
136
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallModelTest::test_flash_attn_2_fp32_ln",
137
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
138
+ ],
139
+ "multi-gpu": [
140
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallModelTest::test_flash_attn_2_fp32_ln",
141
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
142
+ ]
143
+ },
144
+ "blt": {
145
+ "single-gpu": [
146
+ "tests/models/blt/test_modeling_blt.py::BltModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break"
147
+ ],
148
+ "multi-gpu": [
149
+ "tests/models/blt/test_modeling_blt.py::BltModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break"
150
+ ]
151
+ },
152
+ "chameleon": {
153
+ "single-gpu": [
154
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
155
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonModelTest::test_flash_attn_2_fp32_ln",
156
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonVision2SeqModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs",
157
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonVision2SeqModelTest::test_flash_attn_2_fp32_ln"
158
+ ],
159
+ "multi-gpu": [
160
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
161
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonModelTest::test_flash_attn_2_fp32_ln",
162
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonVision2SeqModelTest::test_flash_attn_2_fp32_ln"
163
+ ]
164
+ },
165
+ "clap": {
166
+ "single-gpu": [
167
+ "tests/models/clap/test_modeling_clap.py::ClapAudioModelTest::test_flash_attn_2_inference_equivalence",
168
+ "tests/models/clap/test_modeling_clap.py::ClapAudioModelTest::test_flash_attn_2_inference_equivalence_right_padding",
169
+ "tests/models/clap/test_modeling_clap.py::ClapModelTest::test_flash_attn_2_inference_equivalence",
170
+ "tests/models/clap/test_modeling_clap.py::ClapModelTest::test_flash_attn_2_inference_equivalence_right_padding"
171
+ ],
172
+ "multi-gpu": [
173
+ "tests/models/clap/test_modeling_clap.py::ClapAudioModelTest::test_flash_attn_2_inference_equivalence",
174
+ "tests/models/clap/test_modeling_clap.py::ClapAudioModelTest::test_flash_attn_2_inference_equivalence_right_padding",
175
+ "tests/models/clap/test_modeling_clap.py::ClapModelTest::test_flash_attn_2_inference_equivalence",
176
+ "tests/models/clap/test_modeling_clap.py::ClapModelTest::test_flash_attn_2_inference_equivalence_right_padding"
177
+ ]
178
+ },
179
+ "cohere": {
180
+ "single-gpu": [
181
+ "tests/models/cohere/test_modeling_cohere.py::CohereModelTest::test_flash_attn_2_fp32_ln"
182
+ ],
183
+ "multi-gpu": [
184
+ "tests/models/cohere/test_modeling_cohere.py::CohereModelTest::test_flash_attn_2_fp32_ln"
185
+ ]
186
+ },
187
+ "cohere2": {
188
+ "single-gpu": [
189
+ "tests/models/cohere2/test_modeling_cohere2.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
190
+ ],
191
+ "multi-gpu": [
192
+ "tests/models/cohere2/test_modeling_cohere2.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
193
+ ]
194
+ },
195
+ "cohere2_vision": {
196
+ "single-gpu": [
197
+ "tests/models/cohere2_vision/test_modeling_cohere2_vision.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
198
+ ],
199
+ "multi-gpu": [
200
+ "tests/models/cohere2_vision/test_modeling_cohere2_vision.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
201
+ ]
202
+ },
203
+ "colpali": {
204
+ "single-gpu": [
205
+ "tests/models/colpali/test_modeling_colpali.py::ColPaliForRetrievalModelTest::test_flash_attn_2_inference_equivalence",
206
+ "tests/models/colpali/test_modeling_colpali.py::ColPaliForRetrievalModelTest::test_flash_attn_2_inference_equivalence_right_padding"
207
+ ],
208
+ "multi-gpu": [
209
+ "tests/models/colpali/test_modeling_colpali.py::ColPaliForRetrievalModelTest::test_flash_attn_2_inference_equivalence",
210
+ "tests/models/colpali/test_modeling_colpali.py::ColPaliForRetrievalModelTest::test_flash_attn_2_inference_equivalence_right_padding"
211
+ ]
212
+ },
213
+ "csm": {
214
+ "single-gpu": [
215
+ "tests/models/csm/test_modeling_csm.py::CsmForConditionalGenerationTest::test_flash_attn_2_fp32_ln"
216
+ ],
217
+ "multi-gpu": [
218
+ "tests/models/csm/test_modeling_csm.py::CsmForConditionalGenerationTest::test_flash_attn_2_fp32_ln"
219
+ ]
220
+ },
221
+ "cwm": {
222
+ "single-gpu": [
223
+ "tests/models/cwm/test_modeling_cwm.py::CwmModelTest::test_flash_attn_2_fp32_ln"
224
+ ],
225
+ "multi-gpu": [
226
+ "tests/models/cwm/test_modeling_cwm.py::CwmModelTest::test_flash_attn_2_fp32_ln"
227
+ ]
228
+ },
229
+ "data2vec": {
230
+ "single-gpu": [
231
+ "tests/models/data2vec/test_modeling_data2vec_text.py::Data2VecTextModelTest::test_flash_attn_2_fp32_ln"
232
+ ],
233
+ "multi-gpu": [
234
+ "tests/models/data2vec/test_modeling_data2vec_text.py::Data2VecTextModelTest::test_flash_attn_2_fp32_ln"
235
+ ]
236
+ },
237
+ "dbrx": {
238
+ "single-gpu": [
239
+ "tests/models/dbrx/test_modeling_dbrx.py::DbrxModelTest::test_flash_attn_2_fp32_ln"
240
+ ],
241
+ "multi-gpu": [
242
+ "tests/models/dbrx/test_modeling_dbrx.py::DbrxModelTest::test_flash_attn_2_fp32_ln"
243
+ ]
244
+ },
245
+ "deepseek_v2": {
246
+ "single-gpu": [
247
+ "tests/models/deepseek_v2/test_modeling_deepseek_v2.py::DeepseekV2ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
248
+ "tests/models/deepseek_v2/test_modeling_deepseek_v2.py::DeepseekV2ModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
249
+ "tests/models/deepseek_v2/test_modeling_deepseek_v2.py::DeepseekV2ModelTest::test_flash_attn_2_fp32_ln"
250
+ ],
251
+ "multi-gpu": [
252
+ "tests/models/deepseek_v2/test_modeling_deepseek_v2.py::DeepseekV2ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
253
+ "tests/models/deepseek_v2/test_modeling_deepseek_v2.py::DeepseekV2ModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
254
+ "tests/models/deepseek_v2/test_modeling_deepseek_v2.py::DeepseekV2ModelTest::test_flash_attn_2_fp32_ln"
255
+ ]
256
+ },
257
+ "deepseek_v3": {
258
+ "single-gpu": [
259
+ "tests/models/deepseek_v3/test_modeling_deepseek_v3.py::DeepseekV3ModelTest::test_flash_attn_2_fp32_ln"
260
+ ],
261
+ "multi-gpu": [
262
+ "tests/models/deepseek_v3/test_modeling_deepseek_v3.py::DeepseekV3ModelTest::test_flash_attn_2_fp32_ln"
263
+ ]
264
+ },
265
+ "deepseek_vl": {
266
+ "single-gpu": [
267
+ "tests/models/deepseek_vl/test_modeling_deepseek_vl.py::DeepseekVLModelTest::test_flash_attn_2_fp32_ln"
268
+ ],
269
+ "multi-gpu": [
270
+ "tests/models/deepseek_vl/test_modeling_deepseek_vl.py::DeepseekVLModelTest::test_flash_attn_2_fp32_ln"
271
+ ]
272
+ },
273
+ "deepseek_vl_hybrid": {
274
+ "single-gpu": [
275
+ "tests/models/deepseek_vl_hybrid/test_modeling_deepseek_vl_hybrid.py::DeepseekVLHybridModelTest::test_flash_attention_2_continue_generate_with_position_ids",
276
+ "tests/models/deepseek_vl_hybrid/test_modeling_deepseek_vl_hybrid.py::DeepseekVLHybridModelTest::test_flash_attn_2_fp32_ln",
277
+ "tests/models/deepseek_vl_hybrid/test_modeling_deepseek_vl_hybrid.py::DeepseekVLHybridModelTest::test_flash_attn_2_from_config"
278
+ ],
279
+ "multi-gpu": [
280
+ "tests/models/deepseek_vl_hybrid/test_modeling_deepseek_vl_hybrid.py::DeepseekVLHybridModelTest::test_flash_attention_2_continue_generate_with_position_ids",
281
+ "tests/models/deepseek_vl_hybrid/test_modeling_deepseek_vl_hybrid.py::DeepseekVLHybridModelTest::test_flash_attn_2_fp32_ln",
282
+ "tests/models/deepseek_vl_hybrid/test_modeling_deepseek_vl_hybrid.py::DeepseekVLHybridModelTest::test_flash_attn_2_from_config"
283
+ ]
284
+ },
285
+ "dia": {
286
+ "single-gpu": [
287
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_eager_matches_fa2_generate",
288
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_flash_attn_2_can_dispatch_composite_models",
289
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_flash_attn_2_fp32_ln"
290
+ ],
291
+ "multi-gpu": [
292
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_eager_matches_fa2_generate",
293
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_flash_attn_2_can_dispatch_composite_models",
294
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_flash_attn_2_fp32_ln"
295
+ ]
296
+ },
297
+ "diffllama": {
298
+ "single-gpu": [
299
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_fp32_ln",
300
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_generate_padding_right",
301
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_inference_equivalence",
302
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_inference_equivalence_right_padding"
303
+ ],
304
+ "multi-gpu": [
305
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_fp32_ln",
306
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_generate_padding_right",
307
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_inference_equivalence",
308
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_inference_equivalence_right_padding"
309
+ ]
310
+ },
311
+ "donut": {
312
+ "single-gpu": [
313
+ "tests/models/donut/test_modeling_donut_swin.py::DonutSwinModelTest::test_flash_attn_2_inference_equivalence",
314
+ "tests/models/donut/test_modeling_donut_swin.py::DonutSwinModelTest::test_flash_attn_2_inference_equivalence_right_padding"
315
+ ],
316
+ "multi-gpu": [
317
+ "tests/models/donut/test_modeling_donut_swin.py::DonutSwinModelTest::test_flash_attn_2_inference_equivalence",
318
+ "tests/models/donut/test_modeling_donut_swin.py::DonutSwinModelTest::test_flash_attn_2_inference_equivalence_right_padding"
319
+ ]
320
+ },
321
+ "dots1": {
322
+ "single-gpu": [
323
+ "tests/models/dots1/test_modeling_dots1.py::Dots1ModelTest::test_flash_attn_2_fp32_ln"
324
+ ],
325
+ "multi-gpu": [
326
+ "tests/models/dots1/test_modeling_dots1.py::Dots1ModelTest::test_flash_attn_2_fp32_ln"
327
+ ]
328
+ },
329
+ "efficientloftr": {
330
+ "single-gpu": [
331
+ "tests/models/efficientloftr/test_modeling_efficientloftr.py::EfficientLoFTRModelTest::test_flash_attn_2_inference_equivalence",
332
+ "tests/models/efficientloftr/test_modeling_efficientloftr.py::EfficientLoFTRModelTest::test_flash_attn_2_inference_equivalence_right_padding"
333
+ ],
334
+ "multi-gpu": [
335
+ "tests/models/efficientloftr/test_modeling_efficientloftr.py::EfficientLoFTRModelTest::test_flash_attn_2_inference_equivalence",
336
+ "tests/models/efficientloftr/test_modeling_efficientloftr.py::EfficientLoFTRModelTest::test_flash_attn_2_inference_equivalence_right_padding"
337
+ ]
338
+ },
339
+ "electra": {
340
+ "single-gpu": [
341
+ "tests/models/electra/test_modeling_electra.py::ElectraModelTest::test_flash_attn_2_fp32_ln"
342
+ ],
343
+ "multi-gpu": [
344
+ "tests/models/electra/test_modeling_electra.py::ElectraModelTest::test_flash_attn_2_fp32_ln"
345
+ ]
346
+ },
347
+ "emu3": {
348
+ "single-gpu": [
349
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Text2TextModelTest::test_flash_attn_2_fp32_ln",
350
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_fp32_ln",
351
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_inference_equivalence",
352
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
353
+ ],
354
+ "multi-gpu": [
355
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Text2TextModelTest::test_flash_attn_2_fp32_ln",
356
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_fp32_ln",
357
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_inference_equivalence",
358
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
359
+ ]
360
+ },
361
+ "ernie": {
362
+ "single-gpu": [
363
+ "tests/models/ernie/test_modeling_ernie.py::ErnieModelTest::test_flash_attn_2_fp32_ln"
364
+ ],
365
+ "multi-gpu": [
366
+ "tests/models/ernie/test_modeling_ernie.py::ErnieModelTest::test_flash_attn_2_fp32_ln"
367
+ ]
368
+ },
369
+ "ernie4_5": {
370
+ "single-gpu": [
371
+ "tests/models/ernie4_5/test_modeling_ernie4_5.py::Ernie4_5ModelTest::test_flash_attn_2_fp32_ln"
372
+ ],
373
+ "multi-gpu": [
374
+ "tests/models/ernie4_5/test_modeling_ernie4_5.py::Ernie4_5ModelTest::test_flash_attn_2_fp32_ln"
375
+ ]
376
+ },
377
+ "ernie4_5_moe": {
378
+ "single-gpu": [
379
+ "tests/models/ernie4_5_moe/test_modeling_ernie4_5_moe.py::Ernie4_5_MoeModelTest::test_flash_attn_2_equivalence",
380
+ "tests/models/ernie4_5_moe/test_modeling_ernie4_5_moe.py::Ernie4_5_MoeModelTest::test_flash_attn_2_fp32_ln"
381
+ ],
382
+ "multi-gpu": [
383
+ "tests/models/ernie4_5_moe/test_modeling_ernie4_5_moe.py::Ernie4_5_MoeModelTest::test_flash_attn_2_equivalence",
384
+ "tests/models/ernie4_5_moe/test_modeling_ernie4_5_moe.py::Ernie4_5_MoeModelTest::test_flash_attn_2_fp32_ln"
385
+ ]
386
+ },
387
+ "exaone4": {
388
+ "single-gpu": [
389
+ "tests/models/exaone4/test_modeling_exaone4.py::Exaone4ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
390
+ "tests/models/exaone4/test_modeling_exaone4.py::Exaone4ModelTest::test_flash_attn_2_fp32_ln"
391
+ ],
392
+ "multi-gpu": [
393
+ "tests/models/exaone4/test_modeling_exaone4.py::Exaone4ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
394
+ "tests/models/exaone4/test_modeling_exaone4.py::Exaone4ModelTest::test_flash_attn_2_fp32_ln",
395
+ "tests/models/exaone4/test_modeling_exaone4.py::Exaone4IntegrationTest::test_model_generation_long_flash"
396
+ ]
397
+ },
398
+ "falcon": {
399
+ "single-gpu": [
400
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_fp32_ln",
401
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_inference_equivalence",
402
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_inference_equivalence_right_padding"
403
+ ],
404
+ "multi-gpu": [
405
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_fp32_ln",
406
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_inference_equivalence",
407
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_inference_equivalence_right_padding"
408
+ ]
409
+ },
410
+ "falcon_h1": {
411
+ "single-gpu": [
412
+ "tests/models/falcon_h1/test_modeling_falcon_h1.py::FalconH1ModelTest::test_flash_attn_2_fp32_ln"
413
+ ],
414
+ "multi-gpu": [
415
+ "tests/models/falcon_h1/test_modeling_falcon_h1.py::FalconH1ModelTest::test_flash_attn_2_fp32_ln"
416
+ ]
417
+ },
418
+ "flex_olmo": {
419
+ "single-gpu": [
420
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
421
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
422
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attn_2_equivalence",
423
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attn_2_fp32_ln"
424
+ ],
425
+ "multi-gpu": [
426
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
427
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
428
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attn_2_fp32_ln"
429
+ ]
430
+ },
431
+ "florence2": {
432
+ "single-gpu": [
433
+ "tests/models/florence2/test_modeling_florence2.py::Florence2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
434
+ ],
435
+ "multi-gpu": [
436
+ "tests/models/florence2/test_modeling_florence2.py::Florence2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
437
+ ]
438
+ },
439
+ "fuyu": {
440
+ "single-gpu": [
441
+ "tests/models/fuyu/test_modeling_fuyu.py::FuyuModelTest::test_flash_attn_2_fp32_ln"
442
+ ],
443
+ "multi-gpu": [
444
+ "tests/models/fuyu/test_modeling_fuyu.py::FuyuModelTest::test_flash_attn_2_fp32_ln"
445
+ ]
446
+ },
447
+ "gemma": {
448
+ "single-gpu": [
449
+ "tests/models/gemma/test_modeling_gemma.py::GemmaModelTest::test_flash_attn_2_fp32_ln"
450
+ ],
451
+ "multi-gpu": [
452
+ "tests/models/gemma/test_modeling_gemma.py::GemmaModelTest::test_flash_attn_2_fp32_ln"
453
+ ]
454
+ },
455
+ "gemma2": {
456
+ "single-gpu": [
457
+ "tests/models/gemma2/test_modeling_gemma2.py::Gemma2ModelTest::test_flash_attn_2_fp32_ln"
458
+ ],
459
+ "multi-gpu": [
460
+ "tests/models/gemma2/test_modeling_gemma2.py::Gemma2ModelTest::test_flash_attn_2_fp32_ln"
461
+ ]
462
+ },
463
+ "gemma3": {
464
+ "single-gpu": [
465
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3TextModelTest::test_flash_attn_2_fp32_ln",
466
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3Vision2TextModelTest::test_flash_attn_2_fp32_ln"
467
+ ],
468
+ "multi-gpu": [
469
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3TextModelTest::test_flash_attn_2_fp32_ln",
470
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3Vision2TextModelTest::test_flash_attn_2_fp32_ln"
471
+ ]
472
+ },
473
+ "gemma3n": {
474
+ "single-gpu": [
475
+ "tests/models/gemma3n/test_modeling_gemma3n.py::Gemma3nTextModelTest::test_flash_attn_2_equivalence",
476
+ "tests/models/gemma3n/test_modeling_gemma3n.py::Gemma3nTextModelTest::test_flash_attn_2_inference_equivalence",
477
+ "tests/models/gemma3n/test_modeling_gemma3n.py::Gemma3nTextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
478
+ ],
479
+ "multi-gpu": [
480
+ "tests/models/gemma3n/test_modeling_gemma3n.py::Gemma3nTextModelTest::test_flash_attn_2_equivalence",
481
+ "tests/models/gemma3n/test_modeling_gemma3n.py::Gemma3nTextModelTest::test_flash_attn_2_inference_equivalence",
482
+ "tests/models/gemma3n/test_modeling_gemma3n.py::Gemma3nTextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
483
+ ]
484
+ },
485
+ "glm": {
486
+ "single-gpu": [
487
+ "tests/models/glm/test_modeling_glm.py::GlmModelTest::test_flash_attn_2_fp32_ln"
488
+ ],
489
+ "multi-gpu": [
490
+ "tests/models/glm/test_modeling_glm.py::GlmModelTest::test_flash_attn_2_fp32_ln"
491
+ ]
492
+ },
493
+ "glm4": {
494
+ "single-gpu": [
495
+ "tests/models/glm4/test_modeling_glm4.py::Glm4ModelTest::test_flash_attn_2_equivalence",
496
+ "tests/models/glm4/test_modeling_glm4.py::Glm4ModelTest::test_flash_attn_2_fp32_ln"
497
+ ],
498
+ "multi-gpu": [
499
+ "tests/models/glm4/test_modeling_glm4.py::Glm4ModelTest::test_flash_attn_2_fp32_ln"
500
+ ]
501
+ },
502
+ "glm4_moe": {
503
+ "single-gpu": [
504
+ "tests/models/glm4_moe/test_modeling_glm4_moe.py::Glm4MoeModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
505
+ "tests/models/glm4_moe/test_modeling_glm4_moe.py::Glm4MoeModelTest::test_flash_attn_2_fp32_ln"
506
+ ],
507
+ "multi-gpu": [
508
+ "tests/models/glm4_moe/test_modeling_glm4_moe.py::Glm4MoeModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
509
+ "tests/models/glm4_moe/test_modeling_glm4_moe.py::Glm4MoeModelTest::test_flash_attn_2_fp32_ln"
510
+ ]
511
+ },
512
+ "glm4v": {
513
+ "single-gpu": [
514
+ "tests/models/glm4v/test_modeling_glm4v.py::Glm4vModelTest::test_flash_attn_2_fp32_ln"
515
+ ],
516
+ "multi-gpu": [
517
+ "tests/models/glm4v/test_modeling_glm4v.py::Glm4vModelTest::test_flash_attn_2_fp32_ln"
518
+ ]
519
+ },
520
+ "glm4v_moe": {
521
+ "single-gpu": [
522
+ "tests/models/glm4v_moe/test_modeling_glm4v_moe.py::Glm4vMoeModelTest::test_flash_attn_2_fp32_ln"
523
+ ],
524
+ "multi-gpu": [
525
+ "tests/models/glm4v_moe/test_modeling_glm4v_moe.py::Glm4vMoeModelTest::test_flash_attn_2_fp32_ln"
526
+ ]
527
+ },
528
+ "gpt2": {
529
+ "single-gpu": [
530
+ "tests/models/gpt2/test_modeling_gpt2.py::GPT2ModelTest::test_flash_attn_2_fp32_ln"
531
+ ],
532
+ "multi-gpu": [
533
+ "tests/models/gpt2/test_modeling_gpt2.py::GPT2ModelTest::test_flash_attn_2_fp32_ln"
534
+ ]
535
+ },
536
+ "gpt_bigcode": {
537
+ "single-gpu": [
538
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeModelTest::test_flash_attn_2_fp32_ln",
539
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeMHAModelTest::test_flash_attn_2_fp32_ln"
540
+ ],
541
+ "multi-gpu": [
542
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeModelTest::test_flash_attn_2_fp32_ln",
543
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeMHAModelTest::test_flash_attn_2_fp32_ln"
544
+ ]
545
+ },
546
+ "gpt_neo": {
547
+ "single-gpu": [
548
+ "tests/models/gpt_neo/test_modeling_gpt_neo.py::GPTNeoModelTest::test_eager_matches_fa2_generate",
549
+ "tests/models/gpt_neo/test_modeling_gpt_neo.py::GPTNeoModelTest::test_flash_attn_2_fp32_ln"
550
+ ],
551
+ "multi-gpu": [
552
+ "tests/models/gpt_neo/test_modeling_gpt_neo.py::GPTNeoModelTest::test_flash_attn_2_fp32_ln"
553
+ ]
554
+ },
555
+ "gpt_neox": {
556
+ "single-gpu": [
557
+ "tests/models/gpt_neox/test_modeling_gpt_neox.py::GPTNeoXModelTest::test_flash_attn_2_fp32_ln"
558
+ ],
559
+ "multi-gpu": [
560
+ "tests/models/gpt_neox/test_modeling_gpt_neox.py::GPTNeoXModelTest::test_flash_attn_2_fp32_ln"
561
+ ]
562
+ },
563
+ "gpt_oss": {
564
+ "single-gpu": [
565
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_fp32_ln",
566
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_inference_equivalence",
567
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_inference_equivalence_right_padding"
568
+ ],
569
+ "multi-gpu": [
570
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_fp32_ln",
571
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_inference_equivalence",
572
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_inference_equivalence_right_padding",
573
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs"
574
+ ]
575
+ },
576
+ "gptj": {
577
+ "single-gpu": [
578
+ "tests/models/gptj/test_modeling_gptj.py::GPTJModelTest::test_flash_attn_2_fp32_ln"
579
+ ],
580
+ "multi-gpu": [
581
+ "tests/models/gptj/test_modeling_gptj.py::GPTJModelTest::test_flash_attn_2_fp32_ln"
582
+ ]
583
+ },
584
+ "granite": {
585
+ "single-gpu": [
586
+ "tests/models/granite/test_modeling_granite.py::GraniteModelTest::test_flash_attn_2_fp32_ln"
587
+ ],
588
+ "multi-gpu": [
589
+ "tests/models/granite/test_modeling_granite.py::GraniteModelTest::test_flash_attn_2_fp32_ln"
590
+ ]
591
+ },
592
+ "granite_speech": {
593
+ "single-gpu": [
594
+ "tests/models/granite_speech/test_modeling_granite_speech.py::GraniteSpeechForConditionalGenerationModelTest::test_flash_attn_2_can_dispatch_composite_models"
595
+ ],
596
+ "multi-gpu": [
597
+ "tests/models/granite_speech/test_modeling_granite_speech.py::GraniteSpeechForConditionalGenerationModelTest::test_flash_attn_2_can_dispatch_composite_models"
598
+ ]
599
+ },
600
+ "granitemoe": {
601
+ "single-gpu": [
602
+ "tests/models/granitemoe/test_modeling_granitemoe.py::GraniteMoeModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs",
603
+ "tests/models/granitemoe/test_modeling_granitemoe.py::GraniteMoeModelTest::test_flash_attn_2_fp32_ln"
604
+ ],
605
+ "multi-gpu": [
606
+ "tests/models/granitemoe/test_modeling_granitemoe.py::GraniteMoeModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs",
607
+ "tests/models/granitemoe/test_modeling_granitemoe.py::GraniteMoeModelTest::test_flash_attn_2_fp32_ln"
608
+ ]
609
+ },
610
+ "granitemoehybrid": {
611
+ "single-gpu": [
612
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_fp32_ln",
613
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_inference_equivalence",
614
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_inference_equivalence_right_padding"
615
+ ],
616
+ "multi-gpu": [
617
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_fp32_ln",
618
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_inference_equivalence",
619
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_inference_equivalence_right_padding"
620
+ ]
621
+ },
622
+ "granitemoeshared": {
623
+ "single-gpu": [
624
+ "tests/models/granitemoeshared/test_modeling_granitemoeshared.py::GraniteMoeSharedModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs",
625
+ "tests/models/granitemoeshared/test_modeling_granitemoeshared.py::GraniteMoeSharedModelTest::test_flash_attn_2_fp32_ln"
626
+ ],
627
+ "multi-gpu": [
628
+ "tests/models/granitemoeshared/test_modeling_granitemoeshared.py::GraniteMoeSharedModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs",
629
+ "tests/models/granitemoeshared/test_modeling_granitemoeshared.py::GraniteMoeSharedModelTest::test_flash_attn_2_fp32_ln"
630
+ ]
631
+ },
632
+ "grounding_dino": {
633
+ "single-gpu": [
634
+ "tests/models/grounding_dino/test_modeling_grounding_dino.py::GroundingDinoModelTest::test_flash_attn_2_inference_equivalence",
635
+ "tests/models/grounding_dino/test_modeling_grounding_dino.py::GroundingDinoModelTest::test_flash_attn_2_inference_equivalence_right_padding"
636
+ ],
637
+ "multi-gpu": [
638
+ "tests/models/grounding_dino/test_modeling_grounding_dino.py::GroundingDinoModelTest::test_flash_attn_2_inference_equivalence",
639
+ "tests/models/grounding_dino/test_modeling_grounding_dino.py::GroundingDinoModelTest::test_flash_attn_2_inference_equivalence_right_padding"
640
+ ]
641
+ },
642
+ "helium": {
643
+ "single-gpu": [
644
+ "tests/models/helium/test_modeling_helium.py::HeliumModelTest::test_flash_attn_2_fp32_ln"
645
+ ],
646
+ "multi-gpu": [
647
+ "tests/models/helium/test_modeling_helium.py::HeliumModelTest::test_flash_attn_2_fp32_ln"
648
+ ]
649
+ },
650
+ "hunyuan_v1_dense": {
651
+ "single-gpu": [
652
+ "tests/models/hunyuan_v1_dense/test_modeling_hunyuan_v1_dense.py::HunYuanDenseV1ModelTest::test_flash_attn_2_fp32_ln"
653
+ ],
654
+ "multi-gpu": [
655
+ "tests/models/hunyuan_v1_dense/test_modeling_hunyuan_v1_dense.py::HunYuanDenseV1ModelTest::test_flash_attn_2_fp32_ln"
656
+ ]
657
+ },
658
+ "hunyuan_v1_moe": {
659
+ "single-gpu": [
660
+ "tests/models/hunyuan_v1_moe/test_modeling_hunyuan_v1_moe.py::HunYuanMoEV1ModelTest::test_flash_attn_2_fp32_ln"
661
+ ],
662
+ "multi-gpu": [
663
+ "tests/models/hunyuan_v1_moe/test_modeling_hunyuan_v1_moe.py::HunYuanMoEV1ModelTest::test_flash_attn_2_fp32_ln"
664
+ ]
665
+ },
666
+ "idefics2": {
667
+ "single-gpu": [
668
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
669
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationIntegrationTest::test_flash_attn_2_eager_equivalence"
670
+ ],
671
+ "multi-gpu": [
672
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
673
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationIntegrationTest::test_flash_attn_2_eager_equivalence"
674
+ ]
675
+ },
676
+ "idefics3": {
677
+ "single-gpu": [
678
+ "tests/models/idefics3/test_modeling_idefics3.py::Idefics3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
679
+ ],
680
+ "multi-gpu": [
681
+ "tests/models/idefics3/test_modeling_idefics3.py::Idefics3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
682
+ ]
683
+ },
684
+ "instructblip": {
685
+ "single-gpu": [
686
+ "tests/models/instructblip/test_modeling_instructblip.py::InstructBlipForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_fp32_ln",
687
+ "tests/models/instructblip/test_modeling_instructblip.py::InstructBlipForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_from_config"
688
+ ],
689
+ "multi-gpu": [
690
+ "tests/models/instructblip/test_modeling_instructblip.py::InstructBlipForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_fp32_ln",
691
+ "tests/models/instructblip/test_modeling_instructblip.py::InstructBlipForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_from_config"
692
+ ]
693
+ },
694
+ "instructblipvideo": {
695
+ "single-gpu": [
696
+ "tests/models/instructblipvideo/test_modeling_instructblipvideo.py::InstructBlipVideoForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_fp32_ln",
697
+ "tests/models/instructblipvideo/test_modeling_instructblipvideo.py::InstructBlipVideoForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_from_config"
698
+ ],
699
+ "multi-gpu": [
700
+ "tests/models/instructblipvideo/test_modeling_instructblipvideo.py::InstructBlipVideoForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_fp32_ln",
701
+ "tests/models/instructblipvideo/test_modeling_instructblipvideo.py::InstructBlipVideoForConditionalGenerationDecoderOnlyTest::test_flash_attn_2_from_config"
702
+ ]
703
+ },
704
+ "jamba": {
705
+ "single-gpu": [
706
+ "tests/models/jamba/test_modeling_jamba.py::JambaModelTest::test_flash_attn_2_fp32_ln"
707
+ ],
708
+ "multi-gpu": [
709
+ "tests/models/jamba/test_modeling_jamba.py::JambaModelTest::test_flash_attn_2_fp32_ln"
710
+ ]
711
+ },
712
+ "janus": {
713
+ "single-gpu": [
714
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_fp32_ln",
715
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_inference_equivalence",
716
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
717
+ ],
718
+ "multi-gpu": [
719
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_fp32_ln",
720
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_inference_equivalence",
721
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
722
+ ]
723
+ },
724
+ "jetmoe": {
725
+ "single-gpu": [
726
+ "tests/models/jetmoe/test_modeling_jetmoe.py::JetMoeModelTest::test_flash_attn_2_fp32_ln"
727
+ ],
728
+ "multi-gpu": [
729
+ "tests/models/jetmoe/test_modeling_jetmoe.py::JetMoeModelTest::test_flash_attn_2_fp32_ln"
730
+ ]
731
+ },
732
+ "kosmos2": {
733
+ "single-gpu": [
734
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_eager_matches_fa2_generate",
735
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attention_2_continue_generate_with_position_ids",
736
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_can_dispatch_composite_models",
737
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_fp32_ln",
738
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_from_config",
739
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_inference_equivalence",
740
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_inference_equivalence_right_padding"
741
+ ],
742
+ "multi-gpu": [
743
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_eager_matches_fa2_generate",
744
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attention_2_continue_generate_with_position_ids",
745
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_can_dispatch_composite_models",
746
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_fp32_ln",
747
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_from_config",
748
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_inference_equivalence",
749
+ "tests/models/kosmos2/test_modeling_kosmos2.py::Kosmos2ModelTest::test_flash_attn_2_inference_equivalence_right_padding"
750
+ ]
751
+ },
752
+ "kosmos2_5": {
753
+ "single-gpu": [
754
+ "tests/models/kosmos2_5/test_modeling_kosmos2_5.py::Kosmos2_5ModelTest::test_flash_attn_2_can_dispatch_composite_models"
755
+ ],
756
+ "multi-gpu": [
757
+ "tests/models/kosmos2_5/test_modeling_kosmos2_5.py::Kosmos2_5ModelTest::test_flash_attn_2_can_dispatch_composite_models"
758
+ ]
759
+ },
760
+ "kyutai_speech_to_text": {
761
+ "single-gpu": [
762
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_eager_matches_fa2_generate",
763
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_fp32_ln",
764
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_inference_equivalence",
765
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
766
+ ],
767
+ "multi-gpu": [
768
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_eager_matches_fa2_generate",
769
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_fp32_ln",
770
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_inference_equivalence",
771
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_inference_equivalence_right_padding"
772
+ ]
773
+ },
774
+ "lfm2": {
775
+ "single-gpu": [
776
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
777
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
778
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attn_2_fp32_ln"
779
+ ],
780
+ "multi-gpu": [
781
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
782
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
783
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attn_2_fp32_ln"
784
+ ]
785
+ },
786
+ "lfm2_moe": {
787
+ "single-gpu": [
788
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
789
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
790
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attn_2_fp32_ln"
791
+ ],
792
+ "multi-gpu": [
793
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
794
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
795
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attn_2_fp32_ln"
796
+ ]
797
+ },
798
+ "lfm2_vl": {
799
+ "single-gpu": [
800
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_fp32_ln",
801
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_inference_equivalence",
802
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_inference_equivalence_right_padding"
803
+ ],
804
+ "multi-gpu": [
805
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_fp32_ln",
806
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_inference_equivalence",
807
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_inference_equivalence_right_padding"
808
+ ]
809
+ },
810
+ "llama": {
811
+ "single-gpu": [
812
+ "tests/models/llama/test_modeling_llama.py::LlamaModelTest::test_flash_attn_2_fp32_ln"
813
+ ],
814
+ "multi-gpu": [
815
+ "tests/models/llama/test_modeling_llama.py::LlamaModelTest::test_flash_attn_2_fp32_ln"
816
+ ]
817
+ },
818
+ "llava": {
819
+ "single-gpu": [
820
+ "tests/models/llava/test_modeling_llava.py::LlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
821
+ ],
822
+ "multi-gpu": [
823
+ "tests/models/llava/test_modeling_llava.py::LlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
824
+ ]
825
+ },
826
+ "llava_next": {
827
+ "single-gpu": [
828
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
829
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
830
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
831
+ ],
832
+ "multi-gpu": [
833
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
834
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
835
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
836
+ ]
837
+ },
838
+ "llava_next_video": {
839
+ "single-gpu": [
840
+ "tests/models/llava_next_video/test_modeling_llava_next_video.py::LlavaNextVideoForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
841
+ "tests/models/llava_next_video/test_modeling_llava_next_video.py::LlavaNextVideoForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
842
+ ],
843
+ "multi-gpu": [
844
+ "tests/models/llava_next_video/test_modeling_llava_next_video.py::LlavaNextVideoForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
845
+ "tests/models/llava_next_video/test_modeling_llava_next_video.py::LlavaNextVideoForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
846
+ ]
847
+ },
848
+ "llava_onevision": {
849
+ "single-gpu": [
850
+ "tests/models/llava_onevision/test_modeling_llava_onevision.py::LlavaOnevisionForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
851
+ "tests/models/llava_onevision/test_modeling_llava_onevision.py::LlavaOnevisionForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
852
+ "tests/models/llava_onevision/test_modeling_llava_onevision.py::LlavaOnevisionForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
853
+ ],
854
+ "multi-gpu": [
855
+ "tests/models/llava_onevision/test_modeling_llava_onevision.py::LlavaOnevisionForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
856
+ "tests/models/llava_onevision/test_modeling_llava_onevision.py::LlavaOnevisionForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
857
+ "tests/models/llava_onevision/test_modeling_llava_onevision.py::LlavaOnevisionForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
858
+ ]
859
+ },
860
+ "m2m_100": {
861
+ "single-gpu": [
862
+ "tests/models/m2m_100/test_modeling_m2m_100.py::M2M100ModelTest::test_flash_attn_2_fp32_ln",
863
+ "tests/models/m2m_100/test_modeling_m2m_100.py::M2M100ModelIntegrationTests::test_flash_attn_2_seq_to_seq_generation"
864
+ ],
865
+ "multi-gpu": [
866
+ "tests/models/m2m_100/test_modeling_m2m_100.py::M2M100ModelTest::test_flash_attn_2_fp32_ln",
867
+ "tests/models/m2m_100/test_modeling_m2m_100.py::M2M100ModelIntegrationTests::test_flash_attn_2_seq_to_seq_generation"
868
+ ]
869
+ },
870
+ "marian": {
871
+ "single-gpu": [
872
+ "tests/models/marian/test_modeling_marian.py::MarianModelTest::test_flash_attn_2_fp32_ln",
873
+ "tests/models/marian/test_modeling_marian.py::MarianStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
874
+ ],
875
+ "multi-gpu": [
876
+ "tests/models/marian/test_modeling_marian.py::MarianModelTest::test_flash_attn_2_fp32_ln",
877
+ "tests/models/marian/test_modeling_marian.py::MarianStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
878
+ ]
879
+ },
880
+ "mask2former": {
881
+ "single-gpu": [
882
+ "tests/models/mask2former/test_modeling_mask2former.py::Mask2FormerModelTest::test_flash_attn_2_inference_equivalence",
883
+ "tests/models/mask2former/test_modeling_mask2former.py::Mask2FormerModelTest::test_flash_attn_2_inference_equivalence_right_padding"
884
+ ],
885
+ "multi-gpu": [
886
+ "tests/models/mask2former/test_modeling_mask2former.py::Mask2FormerModelTest::test_flash_attn_2_inference_equivalence",
887
+ "tests/models/mask2former/test_modeling_mask2former.py::Mask2FormerModelTest::test_flash_attn_2_inference_equivalence_right_padding"
888
+ ]
889
+ },
890
+ "maskformer": {
891
+ "single-gpu": [
892
+ "tests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_flash_attn_2_inference_equivalence",
893
+ "tests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_flash_attn_2_inference_equivalence_right_padding",
894
+ "tests/models/maskformer/test_modeling_maskformer_swin.py::MaskFormerSwinModelTest::test_flash_attn_2_inference_equivalence",
895
+ "tests/models/maskformer/test_modeling_maskformer_swin.py::MaskFormerSwinModelTest::test_flash_attn_2_inference_equivalence_right_padding"
896
+ ],
897
+ "multi-gpu": [
898
+ "tests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_flash_attn_2_inference_equivalence",
899
+ "tests/models/maskformer/test_modeling_maskformer.py::MaskFormerModelTest::test_flash_attn_2_inference_equivalence_right_padding",
900
+ "tests/models/maskformer/test_modeling_maskformer_swin.py::MaskFormerSwinModelTest::test_flash_attn_2_inference_equivalence",
901
+ "tests/models/maskformer/test_modeling_maskformer_swin.py::MaskFormerSwinModelTest::test_flash_attn_2_inference_equivalence_right_padding"
902
+ ]
903
+ },
904
+ "mbart": {
905
+ "single-gpu": [
906
+ "tests/models/mbart/test_modeling_mbart.py::MBartModelTest::test_flash_attn_2_fp32_ln",
907
+ "tests/models/mbart/test_modeling_mbart.py::MBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
908
+ ],
909
+ "multi-gpu": [
910
+ "tests/models/mbart/test_modeling_mbart.py::MBartModelTest::test_flash_attn_2_fp32_ln",
911
+ "tests/models/mbart/test_modeling_mbart.py::MBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
912
+ ]
913
+ },
914
+ "minimax": {
915
+ "single-gpu": [
916
+ "tests/models/minimax/test_modeling_minimax.py::MiniMaxModelTest::test_flash_attn_2_fp32_ln"
917
+ ],
918
+ "multi-gpu": [
919
+ "tests/models/minimax/test_modeling_minimax.py::MiniMaxModelTest::test_flash_attn_2_fp32_ln"
920
+ ]
921
+ },
922
+ "ministral": {
923
+ "single-gpu": [
924
+ "tests/models/ministral/test_modeling_ministral.py::MinistralModelTest::test_flash_attn_2_fp32_ln",
925
+ "tests/models/ministral/test_modeling_ministral.py::MinistralIntegrationTest::test_past_sliding_window_generation"
926
+ ],
927
+ "multi-gpu": [
928
+ "tests/models/ministral/test_modeling_ministral.py::MinistralModelTest::test_flash_attn_2_fp32_ln",
929
+ "tests/models/ministral/test_modeling_ministral.py::MinistralIntegrationTest::test_past_sliding_window_generation"
930
+ ]
931
+ },
932
+ "mistral": {
933
+ "single-gpu": [
934
+ "tests/models/mistral/test_modeling_mistral.py::MistralModelTest::test_flash_attn_2_fp32_ln",
935
+ "tests/models/mistral/test_modeling_mistral.py::MistralIntegrationTest::test_model_7b_long_prompt"
936
+ ],
937
+ "multi-gpu": [
938
+ "tests/models/mistral/test_modeling_mistral.py::MistralModelTest::test_flash_attn_2_fp32_ln",
939
+ "tests/models/mistral/test_modeling_mistral.py::MistralIntegrationTest::test_model_7b_long_prompt"
940
+ ]
941
+ },
942
+ "mixtral": {
943
+ "single-gpu": [
944
+ "tests/models/mixtral/test_modeling_mixtral.py::MixtralModelTest::test_flash_attn_2_equivalence",
945
+ "tests/models/mixtral/test_modeling_mixtral.py::MixtralModelTest::test_flash_attn_2_fp32_ln"
946
+ ],
947
+ "multi-gpu": [
948
+ "tests/models/mixtral/test_modeling_mixtral.py::MixtralModelTest::test_flash_attn_2_equivalence",
949
+ "tests/models/mixtral/test_modeling_mixtral.py::MixtralModelTest::test_flash_attn_2_fp32_ln"
950
+ ]
951
+ },
952
+ "mllama": {
953
+ "single-gpu": [
954
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForCausalLMModelTest::test_flash_attn_2_fp32_ln",
955
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_eager_matches_fa2_generate",
956
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
957
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
958
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
959
+ ],
960
+ "multi-gpu": [
961
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForCausalLMModelTest::test_flash_attn_2_fp32_ln",
962
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_eager_matches_fa2_generate",
963
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
964
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
965
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
966
+ ]
967
+ },
968
+ "mm_grounding_dino": {
969
+ "single-gpu": [
970
+ "tests/models/mm_grounding_dino/test_modeling_mm_grounding_dino.py::MMGroundingDinoModelTest::test_flash_attn_2_inference_equivalence",
971
+ "tests/models/mm_grounding_dino/test_modeling_mm_grounding_dino.py::MMGroundingDinoModelTest::test_flash_attn_2_inference_equivalence_right_padding"
972
+ ],
973
+ "multi-gpu": [
974
+ "tests/models/mm_grounding_dino/test_modeling_mm_grounding_dino.py::MMGroundingDinoModelTest::test_flash_attn_2_inference_equivalence",
975
+ "tests/models/mm_grounding_dino/test_modeling_mm_grounding_dino.py::MMGroundingDinoModelTest::test_flash_attn_2_inference_equivalence_right_padding"
976
+ ]
977
+ },
978
+ "modernbert_decoder": {
979
+ "single-gpu": [
980
+ "tests/models/modernbert_decoder/test_modeling_modernbert_decoder.py::ModernBertDecoderModelTest::test_flash_attn_2_fp32_ln"
981
+ ],
982
+ "multi-gpu": [
983
+ "tests/models/modernbert_decoder/test_modeling_modernbert_decoder.py::ModernBertDecoderModelTest::test_flash_attn_2_fp32_ln"
984
+ ]
985
+ },
986
+ "moshi": {
987
+ "single-gpu": [
988
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_fp32_ln",
989
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_inference_equivalence",
990
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_inference_equivalence_right_padding",
991
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_eager_matches_fa2_generate",
992
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_fp32_ln",
993
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_from_config",
994
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_inference_equivalence",
995
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_inference_equivalence_right_padding"
996
+ ],
997
+ "multi-gpu": [
998
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_fp32_ln",
999
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_inference_equivalence",
1000
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_inference_equivalence_right_padding",
1001
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_eager_matches_fa2_generate",
1002
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_fp32_ln",
1003
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_from_config",
1004
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_inference_equivalence",
1005
+ "tests/models/moshi/test_modeling_moshi.py::MoshiTest::test_flash_attn_2_inference_equivalence_right_padding"
1006
+ ]
1007
+ },
1008
+ "musicgen_melody": {
1009
+ "single-gpu": [
1010
+ "tests/models/musicgen_melody/test_modeling_musicgen_melody.py::MusicgenMelodyDecoderTest::test_flash_attn_2_can_dispatch_composite_models",
1011
+ "tests/models/musicgen_melody/test_modeling_musicgen_melody.py::MusicgenMelodyDecoderTest::test_flash_attn_2_inference_equivalence"
1012
+ ],
1013
+ "multi-gpu": [
1014
+ "tests/models/musicgen_melody/test_modeling_musicgen_melody.py::MusicgenMelodyDecoderTest::test_flash_attn_2_can_dispatch_composite_models",
1015
+ "tests/models/musicgen_melody/test_modeling_musicgen_melody.py::MusicgenMelodyDecoderTest::test_flash_attn_2_inference_equivalence"
1016
+ ]
1017
+ },
1018
+ "nemotron": {
1019
+ "single-gpu": [
1020
+ "tests/models/nemotron/test_modeling_nemotron.py::NemotronModelTest::test_flash_attn_2_equivalence",
1021
+ "tests/models/nemotron/test_modeling_nemotron.py::NemotronModelTest::test_flash_attn_2_fp32_ln"
1022
+ ],
1023
+ "multi-gpu": [
1024
+ "tests/models/nemotron/test_modeling_nemotron.py::NemotronModelTest::test_flash_attn_2_equivalence",
1025
+ "tests/models/nemotron/test_modeling_nemotron.py::NemotronModelTest::test_flash_attn_2_fp32_ln"
1026
+ ]
1027
+ },
1028
+ "olmo": {
1029
+ "single-gpu": [
1030
+ "tests/models/olmo/test_modeling_olmo.py::OlmoModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
1031
+ "tests/models/olmo/test_modeling_olmo.py::OlmoModelTest::test_flash_attn_2_fp32_ln"
1032
+ ],
1033
+ "multi-gpu": [
1034
+ "tests/models/olmo/test_modeling_olmo.py::OlmoModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
1035
+ "tests/models/olmo/test_modeling_olmo.py::OlmoModelTest::test_flash_attn_2_fp32_ln"
1036
+ ]
1037
+ },
1038
+ "olmo2": {
1039
+ "single-gpu": [
1040
+ "tests/models/olmo2/test_modeling_olmo2.py::Olmo2ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
1041
+ "tests/models/olmo2/test_modeling_olmo2.py::Olmo2ModelTest::test_flash_attn_2_fp32_ln"
1042
+ ],
1043
+ "multi-gpu": [
1044
+ "tests/models/olmo2/test_modeling_olmo2.py::Olmo2ModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
1045
+ "tests/models/olmo2/test_modeling_olmo2.py::Olmo2ModelTest::test_flash_attn_2_fp32_ln"
1046
+ ]
1047
+ },
1048
+ "olmo3": {
1049
+ "single-gpu": [
1050
+ "tests/models/olmo3/test_modeling_olmo3.py::Olmo3ModelTest::test_flash_attn_2_fp32_ln"
1051
+ ],
1052
+ "multi-gpu": [
1053
+ "tests/models/olmo3/test_modeling_olmo3.py::Olmo3ModelTest::test_flash_attn_2_fp32_ln"
1054
+ ]
1055
+ },
1056
+ "olmoe": {
1057
+ "single-gpu": [
1058
+ "tests/models/olmoe/test_modeling_olmoe.py::OlmoeModelTest::test_flash_attn_2_fp32_ln"
1059
+ ],
1060
+ "multi-gpu": [
1061
+ "tests/models/olmoe/test_modeling_olmoe.py::OlmoeModelTest::test_flash_attn_2_fp32_ln"
1062
+ ]
1063
+ },
1064
+ "omdet_turbo": {
1065
+ "single-gpu": [
1066
+ "tests/models/omdet_turbo/test_modeling_omdet_turbo.py::OmDetTurboModelTest::test_flash_attn_2_inference_equivalence",
1067
+ "tests/models/omdet_turbo/test_modeling_omdet_turbo.py::OmDetTurboModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1068
+ ],
1069
+ "multi-gpu": [
1070
+ "tests/models/omdet_turbo/test_modeling_omdet_turbo.py::OmDetTurboModelTest::test_flash_attn_2_inference_equivalence",
1071
+ "tests/models/omdet_turbo/test_modeling_omdet_turbo.py::OmDetTurboModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1072
+ ]
1073
+ },
1074
+ "oneformer": {
1075
+ "single-gpu": [
1076
+ "tests/models/oneformer/test_modeling_oneformer.py::OneFormerModelTest::test_flash_attn_2_inference_equivalence",
1077
+ "tests/models/oneformer/test_modeling_oneformer.py::OneFormerModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1078
+ ],
1079
+ "multi-gpu": [
1080
+ "tests/models/oneformer/test_modeling_oneformer.py::OneFormerModelTest::test_flash_attn_2_inference_equivalence",
1081
+ "tests/models/oneformer/test_modeling_oneformer.py::OneFormerModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1082
+ ]
1083
+ },
1084
+ "opt": {
1085
+ "single-gpu": [
1086
+ "tests/models/opt/test_modeling_opt.py::OPTModelTest::test_flash_attn_2_fp32_ln"
1087
+ ],
1088
+ "multi-gpu": [
1089
+ "tests/models/opt/test_modeling_opt.py::OPTModelTest::test_flash_attn_2_fp32_ln"
1090
+ ]
1091
+ },
1092
+ "ovis2": {
1093
+ "single-gpu": [
1094
+ "tests/models/ovis2/test_modeling_ovis2.py::Ovis2ModelTest::test_flash_attn_2_fp32_ln"
1095
+ ],
1096
+ "multi-gpu": [
1097
+ "tests/models/ovis2/test_modeling_ovis2.py::Ovis2ModelTest::test_flash_attn_2_fp32_ln"
1098
+ ]
1099
+ },
1100
+ "paligemma": {
1101
+ "single-gpu": [
1102
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
1103
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_from_config",
1104
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
1105
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1106
+ ],
1107
+ "multi-gpu": [
1108
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
1109
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_from_config",
1110
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
1111
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1112
+ ]
1113
+ },
1114
+ "paligemma2": {
1115
+ "single-gpu": [
1116
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
1117
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_from_config",
1118
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
1119
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1120
+ ],
1121
+ "multi-gpu": [
1122
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
1123
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_from_config",
1124
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence",
1125
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1126
+ ]
1127
+ },
1128
+ "pegasus": {
1129
+ "single-gpu": [
1130
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusModelTest::test_flash_attn_2_fp32_ln",
1131
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1132
+ ],
1133
+ "multi-gpu": [
1134
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusModelTest::test_flash_attn_2_fp32_ln",
1135
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1136
+ ]
1137
+ },
1138
+ "pegasus_x": {
1139
+ "single-gpu": [
1140
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_fp32_ln",
1141
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_inference_equivalence",
1142
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_inference_equivalence_right_padding",
1143
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXStandaloneDecoderModelTest::test_flash_attn_2_inference_equivalence",
1144
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXStandaloneDecoderModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1145
+ ],
1146
+ "multi-gpu": [
1147
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_fp32_ln",
1148
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_inference_equivalence",
1149
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_inference_equivalence_right_padding",
1150
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXStandaloneDecoderModelTest::test_flash_attn_2_inference_equivalence",
1151
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXStandaloneDecoderModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1152
+ ]
1153
+ },
1154
+ "perception_lm": {
1155
+ "single-gpu": [
1156
+ "tests/models/perception_lm/test_modeling_perception_lm.py::PerceptionLMForConditionalGenerationModelTest::test_flash_attention_2_continue_generate_with_position_ids"
1157
+ ],
1158
+ "multi-gpu": [
1159
+ "tests/models/perception_lm/test_modeling_perception_lm.py::PerceptionLMForConditionalGenerationModelTest::test_flash_attention_2_continue_generate_with_position_ids"
1160
+ ]
1161
+ },
1162
+ "persimmon": {
1163
+ "single-gpu": [
1164
+ "tests/models/persimmon/test_modeling_persimmon.py::PersimmonModelTest::test_flash_attn_2_fp32_ln"
1165
+ ],
1166
+ "multi-gpu": [
1167
+ "tests/models/persimmon/test_modeling_persimmon.py::PersimmonModelTest::test_flash_attn_2_fp32_ln"
1168
+ ]
1169
+ },
1170
+ "phi": {
1171
+ "single-gpu": [
1172
+ "tests/models/phi/test_modeling_phi.py::PhiModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
1173
+ "tests/models/phi/test_modeling_phi.py::PhiModelTest::test_flash_attn_2_fp32_ln"
1174
+ ],
1175
+ "multi-gpu": [
1176
+ "tests/models/phi/test_modeling_phi.py::PhiModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
1177
+ "tests/models/phi/test_modeling_phi.py::PhiModelTest::test_flash_attn_2_fp32_ln"
1178
+ ]
1179
+ },
1180
+ "phi3": {
1181
+ "single-gpu": [
1182
+ "tests/models/phi3/test_modeling_phi3.py::Phi3ModelTest::test_flash_attn_2_fp32_ln"
1183
+ ],
1184
+ "multi-gpu": [
1185
+ "tests/models/phi3/test_modeling_phi3.py::Phi3ModelTest::test_flash_attn_2_fp32_ln"
1186
+ ]
1187
+ },
1188
+ "phi4_multimodal": {
1189
+ "single-gpu": [
1190
+ "tests/models/phi4_multimodal/test_modeling_phi4_multimodal.py::Phi4MultimodalModelTest::test_flash_attn_2_fp32_ln"
1191
+ ],
1192
+ "multi-gpu": [
1193
+ "tests/models/phi4_multimodal/test_modeling_phi4_multimodal.py::Phi4MultimodalModelTest::test_flash_attn_2_fp32_ln"
1194
+ ]
1195
+ },
1196
+ "phimoe": {
1197
+ "single-gpu": [
1198
+ "tests/models/phimoe/test_modeling_phimoe.py::PhimoeModelTest::test_flash_attn_2_equivalence",
1199
+ "tests/models/phimoe/test_modeling_phimoe.py::PhimoeModelTest::test_flash_attn_2_fp32_ln"
1200
+ ],
1201
+ "multi-gpu": [
1202
+ "tests/models/phimoe/test_modeling_phimoe.py::PhimoeModelTest::test_flash_attn_2_fp32_ln"
1203
+ ]
1204
+ },
1205
+ "pixtral": {
1206
+ "single-gpu": [
1207
+ "tests/models/pixtral/test_modeling_pixtral.py::PixtralVisionModelModelTest::test_flash_attn_2_inference_equivalence",
1208
+ "tests/models/pixtral/test_modeling_pixtral.py::PixtralVisionModelModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1209
+ ],
1210
+ "multi-gpu": [
1211
+ "tests/models/pixtral/test_modeling_pixtral.py::PixtralVisionModelModelTest::test_flash_attn_2_inference_equivalence",
1212
+ "tests/models/pixtral/test_modeling_pixtral.py::PixtralVisionModelModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1213
+ ]
1214
+ },
1215
+ "plbart": {
1216
+ "single-gpu": [
1217
+ "tests/models/plbart/test_modeling_plbart.py::PLBartModelTest::test_flash_attn_2_fp32_ln",
1218
+ "tests/models/plbart/test_modeling_plbart.py::PLBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1219
+ ],
1220
+ "multi-gpu": [
1221
+ "tests/models/plbart/test_modeling_plbart.py::PLBartModelTest::test_flash_attn_2_fp32_ln",
1222
+ "tests/models/plbart/test_modeling_plbart.py::PLBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1223
+ ]
1224
+ },
1225
+ "qwen2": {
1226
+ "single-gpu": [
1227
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2ModelTest::test_flash_attn_2_fp32_ln",
1228
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2IntegrationTest::test_model_450m_long_prompt"
1229
+ ],
1230
+ "multi-gpu": [
1231
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2ModelTest::test_flash_attn_2_fp32_ln",
1232
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2IntegrationTest::test_model_450m_long_prompt"
1233
+ ]
1234
+ },
1235
+ "qwen2_5_omni": {
1236
+ "single-gpu": [
1237
+ "tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py::Qwen2_5OmniThinkerForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
1238
+ "tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py::Qwen2_5OmniModelIntegrationTest::test_small_model_integration_test_batch_flashatt2"
1239
+ ],
1240
+ "multi-gpu": [
1241
+ "tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py::Qwen2_5OmniThinkerForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
1242
+ "tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py::Qwen2_5OmniModelIntegrationTest::test_small_model_integration_test_batch_flashatt2"
1243
+ ]
1244
+ },
1245
+ "qwen2_5_vl": {
1246
+ "single-gpu": [
1247
+ "tests/models/qwen2_5_vl/test_modeling_qwen2_5_vl.py::Qwen2_5_VLModelTest::test_flash_attn_2_fp32_ln",
1248
+ "tests/models/qwen2_5_vl/test_modeling_qwen2_5_vl.py::Qwen2_5_VLIntegrationTest::test_small_model_integration_test_batch_wo_image_flashatt2"
1249
+ ],
1250
+ "multi-gpu": [
1251
+ "tests/models/qwen2_5_vl/test_modeling_qwen2_5_vl.py::Qwen2_5_VLModelTest::test_flash_attn_2_fp32_ln",
1252
+ "tests/models/qwen2_5_vl/test_modeling_qwen2_5_vl.py::Qwen2_5_VLIntegrationTest::test_small_model_integration_test_batch_wo_image_flashatt2"
1253
+ ]
1254
+ },
1255
+ "qwen2_audio": {
1256
+ "single-gpu": [
1257
+ "tests/models/qwen2_audio/test_modeling_qwen2_audio.py::Qwen2AudioForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1258
+ ],
1259
+ "multi-gpu": [
1260
+ "tests/models/qwen2_audio/test_modeling_qwen2_audio.py::Qwen2AudioForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1261
+ ]
1262
+ },
1263
+ "qwen2_moe": {
1264
+ "single-gpu": [
1265
+ "tests/models/qwen2_moe/test_modeling_qwen2_moe.py::Qwen2MoeIntegrationTest::test_model_a2_7b_long_prompt_flash_attn",
1266
+ "tests/models/qwen2_moe/test_modeling_qwen2_moe.py::Qwen2MoeModelTest::test_flash_attn_2_fp32_ln"
1267
+ ],
1268
+ "multi-gpu": [
1269
+ "tests/models/qwen2_moe/test_modeling_qwen2_moe.py::Qwen2MoeIntegrationTest::test_model_a2_7b_long_prompt_flash_attn",
1270
+ "tests/models/qwen2_moe/test_modeling_qwen2_moe.py::Qwen2MoeModelTest::test_flash_attn_2_fp32_ln"
1271
+ ]
1272
+ },
1273
+ "qwen2_vl": {
1274
+ "single-gpu": [
1275
+ "tests/models/qwen2_vl/test_modeling_qwen2_vl.py::Qwen2VLModelTest::test_flash_attn_2_fp32_ln"
1276
+ ],
1277
+ "multi-gpu": [
1278
+ "tests/models/qwen2_vl/test_modeling_qwen2_vl.py::Qwen2VLModelTest::test_flash_attn_2_fp32_ln"
1279
+ ]
1280
+ },
1281
+ "qwen3": {
1282
+ "single-gpu": [
1283
+ "tests/models/qwen3/test_modeling_qwen3.py::Qwen3ModelTest::test_flash_attn_2_fp32_ln",
1284
+ "tests/models/qwen3/test_modeling_qwen3.py::Qwen3IntegrationTest::test_model_600m_long_prompt"
1285
+ ],
1286
+ "multi-gpu": [
1287
+ "tests/models/qwen3/test_modeling_qwen3.py::Qwen3ModelTest::test_flash_attn_2_fp32_ln",
1288
+ "tests/models/qwen3/test_modeling_qwen3.py::Qwen3IntegrationTest::test_model_600m_long_prompt"
1289
+ ]
1290
+ },
1291
+ "qwen3_moe": {
1292
+ "single-gpu": [
1293
+ "tests/models/qwen3_moe/test_modeling_qwen3_moe.py::Qwen3MoeModelTest::test_flash_attn_2_fp32_ln"
1294
+ ],
1295
+ "multi-gpu": [
1296
+ "tests/models/qwen3_moe/test_modeling_qwen3_moe.py::Qwen3MoeModelTest::test_flash_attn_2_fp32_ln",
1297
+ "tests/models/qwen3_moe/test_modeling_qwen3_moe.py::Qwen3MoeIntegrationTest::test_model_15b_a2b_long_prompt"
1298
+ ]
1299
+ },
1300
+ "qwen3_omni_moe": {
1301
+ "single-gpu": [],
1302
+ "multi-gpu": [
1303
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniMoeThinkerForConditionalGenerationModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
1304
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniMoeThinkerForConditionalGenerationModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs",
1305
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniMoeThinkerForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
1306
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniModelIntegrationTest::test_small_model_integration_test_batch_flashatt2"
1307
+ ]
1308
+ },
1309
+ "qwen3_vl": {
1310
+ "single-gpu": [
1311
+ "tests/models/qwen3_vl/test_modeling_qwen3_vl.py::Qwen3VLModelTest::test_flash_attn_2_fp32_ln"
1312
+ ],
1313
+ "multi-gpu": [
1314
+ "tests/models/qwen3_vl/test_modeling_qwen3_vl.py::Qwen3VLModelTest::test_flash_attn_2_fp32_ln"
1315
+ ]
1316
+ },
1317
+ "qwen3_vl_moe": {
1318
+ "single-gpu": [
1319
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeModelTest::test_flash_attn_2_fp32_ln",
1320
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeIntegrationTest::test_small_model_integration_test_batch_flashatt2",
1321
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeIntegrationTest::test_small_model_integration_test_batch_wo_image_flashatt2"
1322
+ ],
1323
+ "multi-gpu": [
1324
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeModelTest::test_flash_attn_2_fp32_ln",
1325
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeIntegrationTest::test_small_model_integration_test_batch_flashatt2",
1326
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeIntegrationTest::test_small_model_integration_test_batch_wo_image_flashatt2"
1327
+ ]
1328
+ },
1329
+ "roberta": {
1330
+ "single-gpu": [
1331
+ "tests/models/roberta/test_modeling_roberta.py::RobertaModelTest::test_flash_attn_2_fp32_ln",
1332
+ "tests/models/roberta/test_modeling_roberta.py::RobertaModelTest::test_flash_attn_2_inference_equivalence"
1333
+ ],
1334
+ "multi-gpu": [
1335
+ "tests/models/roberta/test_modeling_roberta.py::RobertaModelTest::test_flash_attn_2_fp32_ln"
1336
+ ]
1337
+ },
1338
+ "roberta_prelayernorm": {
1339
+ "single-gpu": [
1340
+ "tests/models/roberta_prelayernorm/test_modeling_roberta_prelayernorm.py::RobertaPreLayerNormModelTest::test_flash_attn_2_fp32_ln"
1341
+ ],
1342
+ "multi-gpu": [
1343
+ "tests/models/roberta_prelayernorm/test_modeling_roberta_prelayernorm.py::RobertaPreLayerNormModelTest::test_flash_attn_2_fp32_ln"
1344
+ ]
1345
+ },
1346
+ "sam2": {
1347
+ "single-gpu": [
1348
+ "tests/models/sam2/test_modeling_sam2.py::Sam2ModelTest::test_flash_attn_2_can_dispatch_composite_models"
1349
+ ],
1350
+ "multi-gpu": [
1351
+ "tests/models/sam2/test_modeling_sam2.py::Sam2ModelTest::test_flash_attn_2_can_dispatch_composite_models"
1352
+ ]
1353
+ },
1354
+ "seed_oss": {
1355
+ "single-gpu": [
1356
+ "tests/models/seed_oss/test_modeling_seed_oss.py::SeedOssModelTest::test_flash_attn_2_fp32_ln"
1357
+ ],
1358
+ "multi-gpu": [
1359
+ "tests/models/seed_oss/test_modeling_seed_oss.py::SeedOssModelTest::test_flash_attn_2_fp32_ln"
1360
+ ]
1361
+ },
1362
+ "smollm3": {
1363
+ "single-gpu": [
1364
+ "tests/models/smollm3/test_modeling_smollm3.py::SmolLM3ModelTest::test_flash_attn_2_fp32_ln",
1365
+ "tests/models/smollm3/test_modeling_smollm3.py::SmolLM3IntegrationTest::test_model_3b_long_prompt"
1366
+ ],
1367
+ "multi-gpu": [
1368
+ "tests/models/smollm3/test_modeling_smollm3.py::SmolLM3ModelTest::test_flash_attn_2_fp32_ln",
1369
+ "tests/models/smollm3/test_modeling_smollm3.py::SmolLM3IntegrationTest::test_model_3b_long_prompt"
1370
+ ]
1371
+ },
1372
+ "smolvlm": {
1373
+ "single-gpu": [
1374
+ "tests/models/smolvlm/test_modeling_smolvlm.py::SmolVLMForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1375
+ ],
1376
+ "multi-gpu": [
1377
+ "tests/models/smolvlm/test_modeling_smolvlm.py::SmolVLMForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1378
+ ]
1379
+ },
1380
+ "speech_to_text": {
1381
+ "single-gpu": [
1382
+ "tests/models/speech_to_text/test_modeling_speech_to_text.py::Speech2TextModelTest::test_flash_attn_2_can_dispatch_composite_models"
1383
+ ],
1384
+ "multi-gpu": [
1385
+ "tests/models/speech_to_text/test_modeling_speech_to_text.py::Speech2TextModelTest::test_flash_attn_2_can_dispatch_composite_models"
1386
+ ]
1387
+ },
1388
+ "squeezebert": {
1389
+ "single-gpu": [
1390
+ "tests/models/squeezebert/test_modeling_squeezebert.py::SqueezeBertModelTest::test_flash_attn_2_inference_equivalence",
1391
+ "tests/models/squeezebert/test_modeling_squeezebert.py::SqueezeBertModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1392
+ ],
1393
+ "multi-gpu": [
1394
+ "tests/models/squeezebert/test_modeling_squeezebert.py::SqueezeBertModelTest::test_flash_attn_2_inference_equivalence",
1395
+ "tests/models/squeezebert/test_modeling_squeezebert.py::SqueezeBertModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1396
+ ]
1397
+ },
1398
+ "stablelm": {
1399
+ "single-gpu": [
1400
+ "tests/models/stablelm/test_modeling_stablelm.py::StableLmModelTest::test_flash_attn_2_fp32_ln",
1401
+ "tests/models/stablelm/test_modeling_stablelm.py::StableLmModelIntegrationTest::test_model_3b_long_prompt"
1402
+ ],
1403
+ "multi-gpu": [
1404
+ "tests/models/stablelm/test_modeling_stablelm.py::StableLmModelTest::test_flash_attn_2_fp32_ln",
1405
+ "tests/models/stablelm/test_modeling_stablelm.py::StableLmModelIntegrationTest::test_model_3b_long_prompt"
1406
+ ]
1407
+ },
1408
+ "starcoder2": {
1409
+ "single-gpu": [
1410
+ "tests/models/starcoder2/test_modeling_starcoder2.py::Starcoder2ModelTest::test_flash_attn_2_fp32_ln",
1411
+ "tests/models/starcoder2/test_modeling_starcoder2.py::Starcoder2IntegrationTest::test_starcoder2_batched_generation_fa2"
1412
+ ],
1413
+ "multi-gpu": [
1414
+ "tests/models/starcoder2/test_modeling_starcoder2.py::Starcoder2ModelTest::test_flash_attn_2_fp32_ln",
1415
+ "tests/models/starcoder2/test_modeling_starcoder2.py::Starcoder2IntegrationTest::test_starcoder2_batched_generation_fa2"
1416
+ ]
1417
+ },
1418
+ "swin": {
1419
+ "single-gpu": [
1420
+ "tests/models/swin/test_modeling_swin.py::SwinModelTest::test_flash_attn_2_inference_equivalence",
1421
+ "tests/models/swin/test_modeling_swin.py::SwinModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1422
+ ],
1423
+ "multi-gpu": [
1424
+ "tests/models/swin/test_modeling_swin.py::SwinModelTest::test_flash_attn_2_inference_equivalence",
1425
+ "tests/models/swin/test_modeling_swin.py::SwinModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1426
+ ]
1427
+ },
1428
+ "swin2sr": {
1429
+ "single-gpu": [
1430
+ "tests/models/swin2sr/test_modeling_swin2sr.py::Swin2SRModelTest::test_flash_attn_2_inference_equivalence",
1431
+ "tests/models/swin2sr/test_modeling_swin2sr.py::Swin2SRModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1432
+ ],
1433
+ "multi-gpu": [
1434
+ "tests/models/swin2sr/test_modeling_swin2sr.py::Swin2SRModelTest::test_flash_attn_2_inference_equivalence",
1435
+ "tests/models/swin2sr/test_modeling_swin2sr.py::Swin2SRModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1436
+ ]
1437
+ },
1438
+ "swinv2": {
1439
+ "single-gpu": [
1440
+ "tests/models/swinv2/test_modeling_swinv2.py::Swinv2ModelTest::test_flash_attn_2_inference_equivalence",
1441
+ "tests/models/swinv2/test_modeling_swinv2.py::Swinv2ModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1442
+ ],
1443
+ "multi-gpu": [
1444
+ "tests/models/swinv2/test_modeling_swinv2.py::Swinv2ModelTest::test_flash_attn_2_inference_equivalence",
1445
+ "tests/models/swinv2/test_modeling_swinv2.py::Swinv2ModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1446
+ ]
1447
+ },
1448
+ "t5gemma": {
1449
+ "single-gpu": [
1450
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
1451
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_fp32_ln",
1452
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_inference_equivalence",
1453
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1454
+ ],
1455
+ "multi-gpu": [
1456
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_can_compile_with_attention_mask_None_without_graph_break",
1457
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_fp32_ln",
1458
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_inference_equivalence",
1459
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_inference_equivalence_right_padding"
1460
+ ]
1461
+ },
1462
+ "vaultgemma": {
1463
+ "single-gpu": [
1464
+ "tests/models/vaultgemma/test_modeling_vaultgemma.py::VaultGemmaModelTest::test_flash_attn_2_fp32_ln"
1465
+ ],
1466
+ "multi-gpu": [
1467
+ "tests/models/vaultgemma/test_modeling_vaultgemma.py::VaultGemmaModelTest::test_flash_attn_2_fp32_ln"
1468
+ ]
1469
+ },
1470
+ "video_llama_3": {
1471
+ "single-gpu": [
1472
+ "tests/models/video_llama_3/test_modeling_video_llama_3.py::VideoLlama3ModelTest::test_flash_attn_2_fp32_ln"
1473
+ ],
1474
+ "multi-gpu": [
1475
+ "tests/models/video_llama_3/test_modeling_video_llama_3.py::VideoLlama3ModelTest::test_flash_attn_2_fp32_ln"
1476
+ ]
1477
+ },
1478
+ "video_llava": {
1479
+ "single-gpu": [
1480
+ "tests/models/video_llava/test_modeling_video_llava.py::VideoLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1481
+ ],
1482
+ "multi-gpu": [
1483
+ "tests/models/video_llava/test_modeling_video_llava.py::VideoLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1484
+ ]
1485
+ },
1486
+ "vipllava": {
1487
+ "single-gpu": [
1488
+ "tests/models/vipllava/test_modeling_vipllava.py::VipLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1489
+ ],
1490
+ "multi-gpu": [
1491
+ "tests/models/vipllava/test_modeling_vipllava.py::VipLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1492
+ ]
1493
+ },
1494
+ "voxtral": {
1495
+ "single-gpu": [
1496
+ "tests/models/voxtral/test_modeling_voxtral.py::VoxtralForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1497
+ ],
1498
+ "multi-gpu": [
1499
+ "tests/models/voxtral/test_modeling_voxtral.py::VoxtralForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
1500
+ ]
1501
+ },
1502
+ "wav2vec2": {
1503
+ "single-gpu": [
1504
+ "tests/models/wav2vec2/test_modeling_wav2vec2.py::Wav2Vec2ModelTest::test_flash_attn_2_can_dispatch_composite_models"
1505
+ ],
1506
+ "multi-gpu": [
1507
+ "tests/models/wav2vec2/test_modeling_wav2vec2.py::Wav2Vec2ModelTest::test_flash_attn_2_can_dispatch_composite_models"
1508
+ ]
1509
+ },
1510
+ "whisper": {
1511
+ "single-gpu": [
1512
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_eager_matches_fa2_generate",
1513
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_flash_attn_2_can_dispatch_composite_models",
1514
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_flash_attn_2_fp32_ln",
1515
+ "tests/models/whisper/test_modeling_whisper.py::WhisperStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1516
+ ],
1517
+ "multi-gpu": [
1518
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_eager_matches_fa2_generate",
1519
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_flash_attn_2_can_dispatch_composite_models",
1520
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_flash_attn_2_fp32_ln",
1521
+ "tests/models/whisper/test_modeling_whisper.py::WhisperStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1522
+ ]
1523
+ },
1524
+ "xlm_roberta_xl": {
1525
+ "single-gpu": [
1526
+ "tests/models/xlm_roberta_xl/test_modeling_xlm_roberta_xl.py::XLMRobertaXLModelTest::test_flash_attn_2_fp32_ln"
1527
+ ],
1528
+ "multi-gpu": [
1529
+ "tests/models/xlm_roberta_xl/test_modeling_xlm_roberta_xl.py::XLMRobertaXLModelTest::test_flash_attn_2_fp32_ln"
1530
+ ]
1531
+ },
1532
+ "xmod": {
1533
+ "single-gpu": [
1534
+ "tests/models/xmod/test_modeling_xmod.py::XmodModelTest::test_flash_attn_2_fp32_ln"
1535
+ ],
1536
+ "multi-gpu": [
1537
+ "tests/models/xmod/test_modeling_xmod.py::XmodModelTest::test_flash_attn_2_fp32_ln",
1538
+ "tests/models/xmod/test_modeling_xmod.py::XmodModelTest::test_flash_attn_2_inference_equivalence"
1539
+ ]
1540
+ },
1541
+ "zamba": {
1542
+ "single-gpu": [
1543
+ "tests/models/zamba/test_modeling_zamba.py::ZambaModelTest::test_flash_attn_2_fp32_ln"
1544
+ ],
1545
+ "multi-gpu": [
1546
+ "tests/models/zamba/test_modeling_zamba.py::ZambaModelTest::test_flash_attn_2_fp32_ln"
1547
+ ]
1548
+ },
1549
+ "zamba2": {
1550
+ "single-gpu": [
1551
+ "tests/models/zamba2/test_modeling_zamba2.py::Zamba2ModelTest::test_flash_attn_2_fp32_ln"
1552
+ ],
1553
+ "multi-gpu": [
1554
+ "tests/models/zamba2/test_modeling_zamba2.py::Zamba2ModelTest::test_flash_attn_2_fp32_ln"
1555
+ ]
1556
+ }
1557
+ }