hf-transformers-bot commited on
Commit
19c0cb7
·
verified ·
1 Parent(s): c0418d1

Upload 2025-11-21/ci_results_run_models_gpu/new_failures.json with huggingface_hub

Browse files
2025-11-21/ci_results_run_models_gpu/new_failures.json ADDED
@@ -0,0 +1,1035 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "audioflamingo3": {
27
+ "single-gpu": [
28
+ "tests/models/audioflamingo3/test_modeling_audioflamingo3.py::AudioFlamingo3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
29
+ ],
30
+ "multi-gpu": [
31
+ "tests/models/audioflamingo3/test_modeling_audioflamingo3.py::AudioFlamingo3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
32
+ ]
33
+ },
34
+ "aya_vision": {
35
+ "single-gpu": [
36
+ "tests/models/aya_vision/test_modeling_aya_vision.py::AyaVisionModelTest::test_flash_attn_2_fp32_ln"
37
+ ],
38
+ "multi-gpu": [
39
+ "tests/models/aya_vision/test_modeling_aya_vision.py::AyaVisionModelTest::test_flash_attn_2_fp32_ln"
40
+ ]
41
+ },
42
+ "bamba": {
43
+ "single-gpu": [
44
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_fp32_ln"
45
+ ],
46
+ "multi-gpu": [
47
+ "tests/models/bamba/test_modeling_bamba.py::BambaModelTest::test_flash_attn_2_fp32_ln"
48
+ ]
49
+ },
50
+ "bark": {
51
+ "single-gpu": [
52
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_flash_attn_2_fp32_ln",
53
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_flash_attn_2_fp32_ln"
54
+ ],
55
+ "multi-gpu": [
56
+ "tests/models/bark/test_modeling_bark.py::BarkSemanticModelTest::test_flash_attn_2_fp32_ln",
57
+ "tests/models/bark/test_modeling_bark.py::BarkCoarseModelTest::test_flash_attn_2_fp32_ln"
58
+ ]
59
+ },
60
+ "bart": {
61
+ "single-gpu": [
62
+ "tests/models/bart/test_modeling_bart.py::BartModelTest::test_flash_attn_2_fp32_ln",
63
+ "tests/models/bart/test_modeling_bart.py::BartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
64
+ ],
65
+ "multi-gpu": [
66
+ "tests/models/bart/test_modeling_bart.py::BartModelTest::test_flash_attn_2_fp32_ln",
67
+ "tests/models/bart/test_modeling_bart.py::BartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
68
+ ]
69
+ },
70
+ "bert": {
71
+ "single-gpu": [
72
+ "tests/models/bert/test_modeling_bert.py::BertModelTest::test_flash_attn_2_fp32_ln",
73
+ "tests/models/bert/test_modeling_bert.py::BertModelTest::test_flash_attn_2_inference_equivalence_right_padding"
74
+ ],
75
+ "multi-gpu": [
76
+ "tests/models/bert/test_modeling_bert.py::BertModelTest::test_flash_attn_2_fp32_ln",
77
+ "tests/models/bert/test_modeling_bert.py::BertModelTest::test_flash_attn_2_inference_equivalence_right_padding"
78
+ ]
79
+ },
80
+ "bert_generation": {
81
+ "single-gpu": [
82
+ "tests/models/bert_generation/test_modeling_bert_generation.py::BertGenerationEncoderTest::test_flash_attn_2_fp32_ln"
83
+ ],
84
+ "multi-gpu": [
85
+ "tests/models/bert_generation/test_modeling_bert_generation.py::BertGenerationEncoderTest::test_flash_attn_2_fp32_ln"
86
+ ]
87
+ },
88
+ "biogpt": {
89
+ "single-gpu": [
90
+ "tests/models/biogpt/test_modeling_biogpt.py::BioGptModelTest::test_flash_attn_2_fp32_ln"
91
+ ],
92
+ "multi-gpu": [
93
+ "tests/models/biogpt/test_modeling_biogpt.py::BioGptModelTest::test_flash_attn_2_fp32_ln"
94
+ ]
95
+ },
96
+ "bitnet": {
97
+ "single-gpu": [
98
+ "tests/models/bitnet/test_modeling_bitnet.py::BitNetModelTest::test_flash_attn_2_fp32_ln"
99
+ ],
100
+ "multi-gpu": [
101
+ "tests/models/bitnet/test_modeling_bitnet.py::BitNetModelTest::test_flash_attn_2_fp32_ln"
102
+ ]
103
+ },
104
+ "blenderbot": {
105
+ "single-gpu": [
106
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotModelTest::test_flash_attn_2_fp32_ln",
107
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
108
+ ],
109
+ "multi-gpu": [
110
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotModelTest::test_flash_attn_2_fp32_ln",
111
+ "tests/models/blenderbot/test_modeling_blenderbot.py::BlenderbotStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
112
+ ]
113
+ },
114
+ "blenderbot_small": {
115
+ "single-gpu": [
116
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallModelTest::test_flash_attn_2_fp32_ln",
117
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
118
+ ],
119
+ "multi-gpu": [
120
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallModelTest::test_flash_attn_2_fp32_ln",
121
+ "tests/models/blenderbot_small/test_modeling_blenderbot_small.py::BlenderbotSmallStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
122
+ ]
123
+ },
124
+ "chameleon": {
125
+ "single-gpu": [
126
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonModelTest::test_flash_attn_2_fp32_ln",
127
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonVision2SeqModelTest::test_flash_attn_2_fp32_ln"
128
+ ],
129
+ "multi-gpu": [
130
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonModelTest::test_flash_attn_2_fp32_ln",
131
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonVision2SeqModelTest::test_flash_attn_2_fp32_ln",
132
+ "tests/models/chameleon/test_modeling_chameleon.py::ChameleonVision2SeqModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs"
133
+ ]
134
+ },
135
+ "cohere": {
136
+ "single-gpu": [
137
+ "tests/models/cohere/test_modeling_cohere.py::CohereModelTest::test_flash_attn_2_fp32_ln"
138
+ ],
139
+ "multi-gpu": [
140
+ "tests/models/cohere/test_modeling_cohere.py::CohereModelTest::test_flash_attn_2_fp32_ln"
141
+ ]
142
+ },
143
+ "cohere2": {
144
+ "single-gpu": [
145
+ "tests/models/cohere2/test_modeling_cohere2.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
146
+ ],
147
+ "multi-gpu": [
148
+ "tests/models/cohere2/test_modeling_cohere2.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
149
+ ]
150
+ },
151
+ "cohere2_vision": {
152
+ "single-gpu": [
153
+ "tests/models/cohere2_vision/test_modeling_cohere2_vision.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
154
+ ],
155
+ "multi-gpu": [
156
+ "tests/models/cohere2_vision/test_modeling_cohere2_vision.py::Cohere2ModelTest::test_flash_attn_2_fp32_ln"
157
+ ]
158
+ },
159
+ "csm": {
160
+ "single-gpu": [
161
+ "tests/models/csm/test_modeling_csm.py::CsmForConditionalGenerationTest::test_flash_attn_2_fp32_ln"
162
+ ],
163
+ "multi-gpu": [
164
+ "tests/models/csm/test_modeling_csm.py::CsmForConditionalGenerationTest::test_flash_attn_2_fp32_ln"
165
+ ]
166
+ },
167
+ "cwm": {
168
+ "single-gpu": [
169
+ "tests/models/cwm/test_modeling_cwm.py::CwmModelTest::test_flash_attn_2_fp32_ln"
170
+ ],
171
+ "multi-gpu": [
172
+ "tests/models/cwm/test_modeling_cwm.py::CwmModelTest::test_flash_attn_2_fp32_ln"
173
+ ]
174
+ },
175
+ "data2vec": {
176
+ "single-gpu": [
177
+ "tests/models/data2vec/test_modeling_data2vec_text.py::Data2VecTextModelTest::test_flash_attn_2_fp32_ln"
178
+ ],
179
+ "multi-gpu": [
180
+ "tests/models/data2vec/test_modeling_data2vec_text.py::Data2VecTextModelTest::test_flash_attn_2_fp32_ln"
181
+ ]
182
+ },
183
+ "dbrx": {
184
+ "single-gpu": [
185
+ "tests/models/dbrx/test_modeling_dbrx.py::DbrxModelTest::test_flash_attn_2_fp32_ln"
186
+ ],
187
+ "multi-gpu": [
188
+ "tests/models/dbrx/test_modeling_dbrx.py::DbrxModelTest::test_flash_attn_2_fp32_ln"
189
+ ]
190
+ },
191
+ "deepseek_v3": {
192
+ "single-gpu": [
193
+ "tests/models/deepseek_v3/test_modeling_deepseek_v3.py::DeepseekV3ModelTest::test_flash_attn_2_fp32_ln"
194
+ ],
195
+ "multi-gpu": [
196
+ "tests/models/deepseek_v3/test_modeling_deepseek_v3.py::DeepseekV3ModelTest::test_flash_attn_2_fp32_ln"
197
+ ]
198
+ },
199
+ "deepseek_vl": {
200
+ "single-gpu": [
201
+ "tests/models/deepseek_vl/test_modeling_deepseek_vl.py::DeepseekVLModelTest::test_flash_attn_2_fp32_ln"
202
+ ],
203
+ "multi-gpu": [
204
+ "tests/models/deepseek_vl/test_modeling_deepseek_vl.py::DeepseekVLModelTest::test_flash_attn_2_fp32_ln"
205
+ ]
206
+ },
207
+ "dia": {
208
+ "single-gpu": [
209
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_flash_attn_2_fp32_ln"
210
+ ],
211
+ "multi-gpu": [
212
+ "tests/models/dia/test_modeling_dia.py::DiaModelTest::test_flash_attn_2_fp32_ln"
213
+ ]
214
+ },
215
+ "diffllama": {
216
+ "single-gpu": [
217
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_fp32_ln"
218
+ ],
219
+ "multi-gpu": [
220
+ "tests/models/diffllama/test_modeling_diffllama.py::DiffLlamaModelTest::test_flash_attn_2_fp32_ln"
221
+ ]
222
+ },
223
+ "dots1": {
224
+ "single-gpu": [
225
+ "tests/models/dots1/test_modeling_dots1.py::Dots1ModelTest::test_flash_attn_2_fp32_ln"
226
+ ],
227
+ "multi-gpu": [
228
+ "tests/models/dots1/test_modeling_dots1.py::Dots1ModelTest::test_flash_attn_2_fp32_ln"
229
+ ]
230
+ },
231
+ "electra": {
232
+ "single-gpu": [
233
+ "tests/models/electra/test_modeling_electra.py::ElectraModelTest::test_flash_attn_2_fp32_ln",
234
+ "tests/models/electra/test_modeling_electra.py::ElectraModelTest::test_flash_attn_2_inference_equivalence_right_padding"
235
+ ],
236
+ "multi-gpu": [
237
+ "tests/models/electra/test_modeling_electra.py::ElectraModelTest::test_flash_attn_2_fp32_ln"
238
+ ]
239
+ },
240
+ "emu3": {
241
+ "single-gpu": [
242
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Text2TextModelTest::test_flash_attn_2_fp32_ln",
243
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_fp32_ln"
244
+ ],
245
+ "multi-gpu": [
246
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Text2TextModelTest::test_flash_attn_2_fp32_ln",
247
+ "tests/models/emu3/test_modeling_emu3.py::Emu3Vision2TextModelTest::test_flash_attn_2_fp32_ln"
248
+ ]
249
+ },
250
+ "ernie": {
251
+ "single-gpu": [
252
+ "tests/models/ernie/test_modeling_ernie.py::ErnieModelTest::test_flash_attn_2_fp32_ln"
253
+ ],
254
+ "multi-gpu": [
255
+ "tests/models/ernie/test_modeling_ernie.py::ErnieModelTest::test_flash_attn_2_fp32_ln",
256
+ "tests/models/ernie/test_modeling_ernie.py::ErnieModelTest::test_flash_attn_2_inference_equivalence_right_padding"
257
+ ]
258
+ },
259
+ "ernie4_5": {
260
+ "single-gpu": [
261
+ "tests/models/ernie4_5/test_modeling_ernie4_5.py::Ernie4_5ModelTest::test_flash_attn_2_fp32_ln"
262
+ ],
263
+ "multi-gpu": [
264
+ "tests/models/ernie4_5/test_modeling_ernie4_5.py::Ernie4_5ModelTest::test_flash_attn_2_fp32_ln"
265
+ ]
266
+ },
267
+ "ernie4_5_moe": {
268
+ "single-gpu": [
269
+ "tests/models/ernie4_5_moe/test_modeling_ernie4_5_moe.py::Ernie4_5_MoeModelTest::test_flash_attn_2_fp32_ln"
270
+ ],
271
+ "multi-gpu": [
272
+ "tests/models/ernie4_5_moe/test_modeling_ernie4_5_moe.py::Ernie4_5_MoeModelTest::test_flash_attn_2_fp32_ln"
273
+ ]
274
+ },
275
+ "exaone4": {
276
+ "single-gpu": [
277
+ "tests/models/exaone4/test_modeling_exaone4.py::Exaone4ModelTest::test_flash_attn_2_fp32_ln"
278
+ ],
279
+ "multi-gpu": [
280
+ "tests/models/exaone4/test_modeling_exaone4.py::Exaone4ModelTest::test_flash_attn_2_fp32_ln"
281
+ ]
282
+ },
283
+ "falcon": {
284
+ "single-gpu": [
285
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_fp32_ln"
286
+ ],
287
+ "multi-gpu": [
288
+ "tests/models/falcon/test_modeling_falcon.py::FalconModelTest::test_flash_attn_2_fp32_ln"
289
+ ]
290
+ },
291
+ "falcon_h1": {
292
+ "single-gpu": [
293
+ "tests/models/falcon_h1/test_modeling_falcon_h1.py::FalconH1ModelTest::test_flash_attn_2_fp32_ln"
294
+ ],
295
+ "multi-gpu": [
296
+ "tests/models/falcon_h1/test_modeling_falcon_h1.py::FalconH1ModelTest::test_flash_attn_2_fp32_ln"
297
+ ]
298
+ },
299
+ "flex_olmo": {
300
+ "single-gpu": [
301
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attn_2_fp32_ln"
302
+ ],
303
+ "multi-gpu": [
304
+ "tests/models/flex_olmo/test_modeling_flex_olmo.py::FlexOlmoModelTest::test_flash_attn_2_fp32_ln"
305
+ ]
306
+ },
307
+ "florence2": {
308
+ "single-gpu": [
309
+ "tests/models/florence2/test_modeling_florence2.py::Florence2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
310
+ ],
311
+ "multi-gpu": [
312
+ "tests/models/florence2/test_modeling_florence2.py::Florence2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
313
+ ]
314
+ },
315
+ "gemma": {
316
+ "single-gpu": [
317
+ "tests/models/gemma/test_modeling_gemma.py::GemmaModelTest::test_flash_attn_2_fp32_ln"
318
+ ],
319
+ "multi-gpu": [
320
+ "tests/models/gemma/test_modeling_gemma.py::GemmaModelTest::test_flash_attn_2_fp32_ln"
321
+ ]
322
+ },
323
+ "gemma2": {
324
+ "single-gpu": [
325
+ "tests/models/gemma2/test_modeling_gemma2.py::Gemma2ModelTest::test_flash_attn_2_fp32_ln"
326
+ ],
327
+ "multi-gpu": [
328
+ "tests/models/gemma2/test_modeling_gemma2.py::Gemma2ModelTest::test_flash_attn_2_fp32_ln"
329
+ ]
330
+ },
331
+ "gemma3": {
332
+ "single-gpu": [
333
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3TextModelTest::test_flash_attn_2_fp32_ln",
334
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3Vision2TextModelTest::test_flash_attn_2_fp32_ln"
335
+ ],
336
+ "multi-gpu": [
337
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3TextModelTest::test_flash_attn_2_fp32_ln",
338
+ "tests/models/gemma3/test_modeling_gemma3.py::Gemma3Vision2TextModelTest::test_flash_attn_2_fp32_ln"
339
+ ]
340
+ },
341
+ "glm": {
342
+ "single-gpu": [
343
+ "tests/models/glm/test_modeling_glm.py::GlmModelTest::test_flash_attn_2_fp32_ln"
344
+ ],
345
+ "multi-gpu": [
346
+ "tests/models/glm/test_modeling_glm.py::GlmModelTest::test_flash_attn_2_fp32_ln"
347
+ ]
348
+ },
349
+ "glm4": {
350
+ "single-gpu": [
351
+ "tests/models/glm4/test_modeling_glm4.py::Glm4ModelTest::test_flash_attn_2_fp32_ln"
352
+ ],
353
+ "multi-gpu": [
354
+ "tests/models/glm4/test_modeling_glm4.py::Glm4ModelTest::test_flash_attn_2_fp32_ln"
355
+ ]
356
+ },
357
+ "glm46v": {
358
+ "single-gpu": [
359
+ "tests/models/glm46v/test_modeling_glm46v.py::Glm46VModelTest::test_flash_attn_2_fp32_ln"
360
+ ],
361
+ "multi-gpu": [
362
+ "tests/models/glm46v/test_modeling_glm46v.py::Glm46VModelTest::test_flash_attn_2_fp32_ln"
363
+ ]
364
+ },
365
+ "glm4_moe": {
366
+ "single-gpu": [
367
+ "tests/models/glm4_moe/test_modeling_glm4_moe.py::Glm4MoeModelTest::test_flash_attn_2_fp32_ln"
368
+ ],
369
+ "multi-gpu": [
370
+ "tests/models/glm4_moe/test_modeling_glm4_moe.py::Glm4MoeModelTest::test_flash_attn_2_fp32_ln"
371
+ ]
372
+ },
373
+ "glm4v": {
374
+ "single-gpu": [
375
+ "tests/models/glm4v/test_modeling_glm4v.py::Glm4vModelTest::test_flash_attn_2_fp32_ln"
376
+ ],
377
+ "multi-gpu": [
378
+ "tests/models/glm4v/test_modeling_glm4v.py::Glm4vModelTest::test_flash_attn_2_fp32_ln"
379
+ ]
380
+ },
381
+ "glm4v_moe": {
382
+ "single-gpu": [
383
+ "tests/models/glm4v_moe/test_modeling_glm4v_moe.py::Glm4vMoeModelTest::test_flash_attn_2_fp32_ln"
384
+ ],
385
+ "multi-gpu": [
386
+ "tests/models/glm4v_moe/test_modeling_glm4v_moe.py::Glm4vMoeModelTest::test_flash_attn_2_fp32_ln"
387
+ ]
388
+ },
389
+ "gpt2": {
390
+ "single-gpu": [
391
+ "tests/models/gpt2/test_modeling_gpt2.py::GPT2ModelTest::test_flash_attn_2_fp32_ln"
392
+ ],
393
+ "multi-gpu": [
394
+ "tests/models/gpt2/test_modeling_gpt2.py::GPT2ModelTest::test_flash_attn_2_fp32_ln"
395
+ ]
396
+ },
397
+ "gpt_bigcode": {
398
+ "single-gpu": [
399
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeModelTest::test_flash_attn_2_fp32_ln",
400
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeMHAModelTest::test_flash_attn_2_fp32_ln"
401
+ ],
402
+ "multi-gpu": [
403
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeModelTest::test_flash_attn_2_fp32_ln",
404
+ "tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py::GPTBigCodeMHAModelTest::test_flash_attn_2_fp32_ln"
405
+ ]
406
+ },
407
+ "gpt_neo": {
408
+ "single-gpu": [
409
+ "tests/models/gpt_neo/test_modeling_gpt_neo.py::GPTNeoModelTest::test_eager_matches_fa2_generate",
410
+ "tests/models/gpt_neo/test_modeling_gpt_neo.py::GPTNeoModelTest::test_flash_attn_2_fp32_ln"
411
+ ],
412
+ "multi-gpu": [
413
+ "tests/models/gpt_neo/test_modeling_gpt_neo.py::GPTNeoModelTest::test_flash_attn_2_fp32_ln"
414
+ ]
415
+ },
416
+ "gpt_neox": {
417
+ "single-gpu": [
418
+ "tests/models/gpt_neox/test_modeling_gpt_neox.py::GPTNeoXModelTest::test_flash_attn_2_fp32_ln"
419
+ ],
420
+ "multi-gpu": [
421
+ "tests/models/gpt_neox/test_modeling_gpt_neox.py::GPTNeoXModelTest::test_flash_attn_2_fp32_ln"
422
+ ]
423
+ },
424
+ "gpt_oss": {
425
+ "single-gpu": [
426
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids_and_fa_kwargs",
427
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_fp32_ln"
428
+ ],
429
+ "multi-gpu": [
430
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attn_2_fp32_ln",
431
+ "tests/models/gpt_oss/test_modeling_gpt_oss.py::GptOssModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids"
432
+ ]
433
+ },
434
+ "gptj": {
435
+ "single-gpu": [
436
+ "tests/models/gptj/test_modeling_gptj.py::GPTJModelTest::test_flash_attn_2_fp32_ln"
437
+ ],
438
+ "multi-gpu": [
439
+ "tests/models/gptj/test_modeling_gptj.py::GPTJModelTest::test_flash_attn_2_fp32_ln"
440
+ ]
441
+ },
442
+ "granite": {
443
+ "single-gpu": [
444
+ "tests/models/granite/test_modeling_granite.py::GraniteModelTest::test_flash_attn_2_fp32_ln"
445
+ ],
446
+ "multi-gpu": [
447
+ "tests/models/granite/test_modeling_granite.py::GraniteModelTest::test_flash_attn_2_fp32_ln"
448
+ ]
449
+ },
450
+ "granitemoe": {
451
+ "single-gpu": [
452
+ "tests/models/granitemoe/test_modeling_granitemoe.py::GraniteMoeModelTest::test_flash_attn_2_fp32_ln"
453
+ ],
454
+ "multi-gpu": [
455
+ "tests/models/granitemoe/test_modeling_granitemoe.py::GraniteMoeModelTest::test_flash_attn_2_fp32_ln"
456
+ ]
457
+ },
458
+ "granitemoehybrid": {
459
+ "single-gpu": [
460
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_fp32_ln"
461
+ ],
462
+ "multi-gpu": [
463
+ "tests/models/granitemoehybrid/test_modeling_granitemoehybrid.py::GraniteMoeHybridModelTest::test_flash_attn_2_fp32_ln"
464
+ ]
465
+ },
466
+ "granitemoeshared": {
467
+ "single-gpu": [
468
+ "tests/models/granitemoeshared/test_modeling_granitemoeshared.py::GraniteMoeSharedModelTest::test_flash_attn_2_fp32_ln"
469
+ ],
470
+ "multi-gpu": [
471
+ "tests/models/granitemoeshared/test_modeling_granitemoeshared.py::GraniteMoeSharedModelTest::test_flash_attn_2_fp32_ln"
472
+ ]
473
+ },
474
+ "helium": {
475
+ "single-gpu": [
476
+ "tests/models/helium/test_modeling_helium.py::HeliumModelTest::test_flash_attn_2_fp32_ln"
477
+ ],
478
+ "multi-gpu": [
479
+ "tests/models/helium/test_modeling_helium.py::HeliumModelTest::test_flash_attn_2_fp32_ln"
480
+ ]
481
+ },
482
+ "hunyuan_v1_dense": {
483
+ "single-gpu": [
484
+ "tests/models/hunyuan_v1_dense/test_modeling_hunyuan_v1_dense.py::HunYuanDenseV1ModelTest::test_flash_attn_2_fp32_ln"
485
+ ],
486
+ "multi-gpu": [
487
+ "tests/models/hunyuan_v1_dense/test_modeling_hunyuan_v1_dense.py::HunYuanDenseV1ModelTest::test_flash_attn_2_fp32_ln"
488
+ ]
489
+ },
490
+ "hunyuan_v1_moe": {
491
+ "single-gpu": [
492
+ "tests/models/hunyuan_v1_moe/test_modeling_hunyuan_v1_moe.py::HunYuanMoEV1ModelTest::test_flash_attn_2_equivalence",
493
+ "tests/models/hunyuan_v1_moe/test_modeling_hunyuan_v1_moe.py::HunYuanMoEV1ModelTest::test_flash_attn_2_fp32_ln"
494
+ ],
495
+ "multi-gpu": [
496
+ "tests/models/hunyuan_v1_moe/test_modeling_hunyuan_v1_moe.py::HunYuanMoEV1ModelTest::test_flash_attn_2_equivalence",
497
+ "tests/models/hunyuan_v1_moe/test_modeling_hunyuan_v1_moe.py::HunYuanMoEV1ModelTest::test_flash_attn_2_fp32_ln"
498
+ ]
499
+ },
500
+ "idefics2": {
501
+ "single-gpu": [
502
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
503
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationIntegrationTest::test_flash_attn_2_eager_equivalence"
504
+ ],
505
+ "multi-gpu": [
506
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
507
+ "tests/models/idefics2/test_modeling_idefics2.py::Idefics2ForConditionalGenerationIntegrationTest::test_flash_attn_2_eager_equivalence"
508
+ ]
509
+ },
510
+ "idefics3": {
511
+ "single-gpu": [
512
+ "tests/models/idefics3/test_modeling_idefics3.py::Idefics3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
513
+ ],
514
+ "multi-gpu": [
515
+ "tests/models/idefics3/test_modeling_idefics3.py::Idefics3ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
516
+ ]
517
+ },
518
+ "jamba": {
519
+ "single-gpu": [
520
+ "tests/models/jamba/test_modeling_jamba.py::JambaModelTest::test_flash_attn_2_fp32_ln"
521
+ ],
522
+ "multi-gpu": [
523
+ "tests/models/jamba/test_modeling_jamba.py::JambaModelTest::test_flash_attn_2_fp32_ln"
524
+ ]
525
+ },
526
+ "janus": {
527
+ "single-gpu": [
528
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_fp32_ln"
529
+ ],
530
+ "multi-gpu": [
531
+ "tests/models/janus/test_modeling_janus.py::JanusVisionText2TextModelTest::test_flash_attn_2_fp32_ln"
532
+ ]
533
+ },
534
+ "jetmoe": {
535
+ "single-gpu": [],
536
+ "multi-gpu": [
537
+ "tests/models/jetmoe/test_modeling_jetmoe.py::JetMoeModelTest::test_flash_attn_2_equivalence"
538
+ ]
539
+ },
540
+ "kyutai_speech_to_text": {
541
+ "single-gpu": [
542
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_fp32_ln"
543
+ ],
544
+ "multi-gpu": [
545
+ "tests/models/kyutai_speech_to_text/test_modeling_kyutai_speech_to_text.py::KyutaiSpeechToTextModelTest::test_flash_attn_2_fp32_ln"
546
+ ]
547
+ },
548
+ "lfm2": {
549
+ "single-gpu": [
550
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attn_2_fp32_ln"
551
+ ],
552
+ "multi-gpu": [
553
+ "tests/models/lfm2/test_modeling_lfm2.py::Lfm2ModelTest::test_flash_attn_2_fp32_ln"
554
+ ]
555
+ },
556
+ "lfm2_moe": {
557
+ "single-gpu": [
558
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attn_2_fp32_ln"
559
+ ],
560
+ "multi-gpu": [
561
+ "tests/models/lfm2_moe/test_modeling_lfm2_moe.py::Lfm2MoeModelTest::test_flash_attn_2_fp32_ln"
562
+ ]
563
+ },
564
+ "lfm2_vl": {
565
+ "single-gpu": [
566
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_fp32_ln"
567
+ ],
568
+ "multi-gpu": [
569
+ "tests/models/lfm2_vl/test_modeling_lfm2_vl.py::Lfm2VlModelTest::test_flash_attn_2_fp32_ln"
570
+ ]
571
+ },
572
+ "llama": {
573
+ "single-gpu": [
574
+ "tests/models/llama/test_modeling_llama.py::LlamaModelTest::test_flash_attn_2_fp32_ln"
575
+ ],
576
+ "multi-gpu": [
577
+ "tests/models/llama/test_modeling_llama.py::LlamaModelTest::test_flash_attn_2_fp32_ln"
578
+ ]
579
+ },
580
+ "llava": {
581
+ "single-gpu": [
582
+ "tests/models/llava/test_modeling_llava.py::LlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
583
+ ],
584
+ "multi-gpu": [
585
+ "tests/models/llava/test_modeling_llava.py::LlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
586
+ ]
587
+ },
588
+ "llava_next": {
589
+ "single-gpu": [
590
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
591
+ ],
592
+ "multi-gpu": [
593
+ "tests/models/llava_next/test_modeling_llava_next.py::LlavaNextForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
594
+ ]
595
+ },
596
+ "m2m_100": {
597
+ "single-gpu": [
598
+ "tests/models/m2m_100/test_modeling_m2m_100.py::M2M100ModelTest::test_flash_attn_2_fp32_ln"
599
+ ],
600
+ "multi-gpu": [
601
+ "tests/models/m2m_100/test_modeling_m2m_100.py::M2M100ModelTest::test_flash_attn_2_fp32_ln"
602
+ ]
603
+ },
604
+ "marian": {
605
+ "single-gpu": [
606
+ "tests/models/marian/test_modeling_marian.py::MarianModelTest::test_flash_attn_2_fp32_ln",
607
+ "tests/models/marian/test_modeling_marian.py::MarianStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
608
+ ],
609
+ "multi-gpu": [
610
+ "tests/models/marian/test_modeling_marian.py::MarianModelTest::test_flash_attn_2_fp32_ln",
611
+ "tests/models/marian/test_modeling_marian.py::MarianStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
612
+ ]
613
+ },
614
+ "mbart": {
615
+ "single-gpu": [
616
+ "tests/models/mbart/test_modeling_mbart.py::MBartModelTest::test_flash_attn_2_fp32_ln",
617
+ "tests/models/mbart/test_modeling_mbart.py::MBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
618
+ ],
619
+ "multi-gpu": [
620
+ "tests/models/mbart/test_modeling_mbart.py::MBartModelTest::test_flash_attn_2_fp32_ln",
621
+ "tests/models/mbart/test_modeling_mbart.py::MBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
622
+ ]
623
+ },
624
+ "minimax": {
625
+ "single-gpu": [
626
+ "tests/models/minimax/test_modeling_minimax.py::MiniMaxModelTest::test_flash_attn_2_fp32_ln"
627
+ ],
628
+ "multi-gpu": [
629
+ "tests/models/minimax/test_modeling_minimax.py::MiniMaxModelTest::test_flash_attn_2_fp32_ln"
630
+ ]
631
+ },
632
+ "ministral": {
633
+ "single-gpu": [
634
+ "tests/models/ministral/test_modeling_ministral.py::MinistralModelTest::test_flash_attn_2_fp32_ln",
635
+ "tests/models/ministral/test_modeling_ministral.py::MinistralIntegrationTest::test_past_sliding_window_generation"
636
+ ],
637
+ "multi-gpu": [
638
+ "tests/models/ministral/test_modeling_ministral.py::MinistralModelTest::test_flash_attn_2_fp32_ln"
639
+ ]
640
+ },
641
+ "mistral": {
642
+ "single-gpu": [
643
+ "tests/models/mistral/test_modeling_mistral.py::MistralModelTest::test_flash_attn_2_fp32_ln",
644
+ "tests/models/mistral/test_modeling_mistral.py::MistralIntegrationTest::test_model_7b_long_prompt"
645
+ ],
646
+ "multi-gpu": [
647
+ "tests/models/mistral/test_modeling_mistral.py::MistralModelTest::test_flash_attn_2_fp32_ln",
648
+ "tests/models/mistral/test_modeling_mistral.py::MistralIntegrationTest::test_model_7b_long_prompt"
649
+ ]
650
+ },
651
+ "mixtral": {
652
+ "single-gpu": [
653
+ "tests/models/mixtral/test_modeling_mixtral.py::MixtralModelTest::test_flash_attn_2_fp32_ln"
654
+ ],
655
+ "multi-gpu": [
656
+ "tests/models/mixtral/test_modeling_mixtral.py::MixtralModelTest::test_flash_attn_2_fp32_ln"
657
+ ]
658
+ },
659
+ "mllama": {
660
+ "single-gpu": [
661
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForCausalLMModelTest::test_flash_attn_2_fp32_ln",
662
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
663
+ ],
664
+ "multi-gpu": [
665
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForCausalLMModelTest::test_flash_attn_2_fp32_ln",
666
+ "tests/models/mllama/test_modeling_mllama.py::MllamaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
667
+ ]
668
+ },
669
+ "modernbert_decoder": {
670
+ "single-gpu": [
671
+ "tests/models/modernbert_decoder/test_modeling_modernbert_decoder.py::ModernBertDecoderModelTest::test_flash_attn_2_fp32_ln"
672
+ ],
673
+ "multi-gpu": [
674
+ "tests/models/modernbert_decoder/test_modeling_modernbert_decoder.py::ModernBertDecoderModelTest::test_flash_attn_2_fp32_ln"
675
+ ]
676
+ },
677
+ "moshi": {
678
+ "single-gpu": [
679
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_fp32_ln"
680
+ ],
681
+ "multi-gpu": [
682
+ "tests/models/moshi/test_modeling_moshi.py::MoshiDecoderTest::test_flash_attn_2_fp32_ln"
683
+ ]
684
+ },
685
+ "nemotron": {
686
+ "single-gpu": [
687
+ "tests/models/nemotron/test_modeling_nemotron.py::NemotronModelTest::test_flash_attn_2_fp32_ln"
688
+ ],
689
+ "multi-gpu": [
690
+ "tests/models/nemotron/test_modeling_nemotron.py::NemotronModelTest::test_flash_attn_2_fp32_ln"
691
+ ]
692
+ },
693
+ "olmo": {
694
+ "single-gpu": [
695
+ "tests/models/olmo/test_modeling_olmo.py::OlmoModelTest::test_flash_attn_2_fp32_ln"
696
+ ],
697
+ "multi-gpu": [
698
+ "tests/models/olmo/test_modeling_olmo.py::OlmoModelTest::test_flash_attn_2_fp32_ln"
699
+ ]
700
+ },
701
+ "olmo2": {
702
+ "single-gpu": [
703
+ "tests/models/olmo2/test_modeling_olmo2.py::Olmo2ModelTest::test_flash_attn_2_fp32_ln"
704
+ ],
705
+ "multi-gpu": [
706
+ "tests/models/olmo2/test_modeling_olmo2.py::Olmo2ModelTest::test_flash_attn_2_fp32_ln"
707
+ ]
708
+ },
709
+ "olmo3": {
710
+ "single-gpu": [
711
+ "tests/models/olmo3/test_modeling_olmo3.py::Olmo3ModelTest::test_flash_attn_2_fp32_ln"
712
+ ],
713
+ "multi-gpu": [
714
+ "tests/models/olmo3/test_modeling_olmo3.py::Olmo3ModelTest::test_flash_attn_2_fp32_ln"
715
+ ]
716
+ },
717
+ "olmoe": {
718
+ "single-gpu": [
719
+ "tests/models/olmoe/test_modeling_olmoe.py::OlmoeModelTest::test_flash_attn_2_fp32_ln"
720
+ ],
721
+ "multi-gpu": [
722
+ "tests/models/olmoe/test_modeling_olmoe.py::OlmoeModelTest::test_flash_attn_2_fp32_ln"
723
+ ]
724
+ },
725
+ "opt": {
726
+ "single-gpu": [
727
+ "tests/models/opt/test_modeling_opt.py::OPTModelTest::test_flash_attn_2_fp32_ln"
728
+ ],
729
+ "multi-gpu": [
730
+ "tests/models/opt/test_modeling_opt.py::OPTModelTest::test_flash_attn_2_fp32_ln"
731
+ ]
732
+ },
733
+ "ovis2": {
734
+ "single-gpu": [
735
+ "tests/models/ovis2/test_modeling_ovis2.py::Ovis2ModelTest::test_flash_attn_2_fp32_ln"
736
+ ],
737
+ "multi-gpu": [
738
+ "tests/models/ovis2/test_modeling_ovis2.py::Ovis2ModelTest::test_flash_attn_2_fp32_ln"
739
+ ]
740
+ },
741
+ "paligemma": {
742
+ "single-gpu": [
743
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
744
+ ],
745
+ "multi-gpu": [
746
+ "tests/models/paligemma/test_modeling_paligemma.py::PaliGemmaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
747
+ ]
748
+ },
749
+ "paligemma2": {
750
+ "single-gpu": [
751
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
752
+ ],
753
+ "multi-gpu": [
754
+ "tests/models/paligemma2/test_modeling_paligemma2.py::PaliGemma2ForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
755
+ ]
756
+ },
757
+ "pegasus": {
758
+ "single-gpu": [
759
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusModelTest::test_flash_attn_2_fp32_ln",
760
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
761
+ ],
762
+ "multi-gpu": [
763
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusModelTest::test_flash_attn_2_fp32_ln",
764
+ "tests/models/pegasus/test_modeling_pegasus.py::PegasusStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
765
+ ]
766
+ },
767
+ "pegasus_x": {
768
+ "single-gpu": [
769
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_fp32_ln"
770
+ ],
771
+ "multi-gpu": [
772
+ "tests/models/pegasus_x/test_modeling_pegasus_x.py::PegasusXModelTest::test_flash_attn_2_fp32_ln"
773
+ ]
774
+ },
775
+ "persimmon": {
776
+ "single-gpu": [
777
+ "tests/models/persimmon/test_modeling_persimmon.py::PersimmonModelTest::test_flash_attn_2_fp32_ln"
778
+ ],
779
+ "multi-gpu": [
780
+ "tests/models/persimmon/test_modeling_persimmon.py::PersimmonModelTest::test_flash_attn_2_fp32_ln"
781
+ ]
782
+ },
783
+ "phi": {
784
+ "single-gpu": [
785
+ "tests/models/phi/test_modeling_phi.py::PhiModelTest::test_flash_attn_2_fp32_ln"
786
+ ],
787
+ "multi-gpu": [
788
+ "tests/models/phi/test_modeling_phi.py::PhiModelTest::test_flash_attn_2_fp32_ln"
789
+ ]
790
+ },
791
+ "phi3": {
792
+ "single-gpu": [
793
+ "tests/models/phi3/test_modeling_phi3.py::Phi3ModelTest::test_flash_attn_2_fp32_ln"
794
+ ],
795
+ "multi-gpu": [
796
+ "tests/models/phi3/test_modeling_phi3.py::Phi3ModelTest::test_flash_attn_2_fp32_ln"
797
+ ]
798
+ },
799
+ "phi4_multimodal": {
800
+ "single-gpu": [
801
+ "tests/models/phi4_multimodal/test_modeling_phi4_multimodal.py::Phi4MultimodalModelTest::test_flash_attn_2_fp32_ln"
802
+ ],
803
+ "multi-gpu": [
804
+ "tests/models/phi4_multimodal/test_modeling_phi4_multimodal.py::Phi4MultimodalModelTest::test_flash_attn_2_fp32_ln"
805
+ ]
806
+ },
807
+ "plbart": {
808
+ "single-gpu": [
809
+ "tests/models/plbart/test_modeling_plbart.py::PLBartModelTest::test_flash_attn_2_fp32_ln",
810
+ "tests/models/plbart/test_modeling_plbart.py::PLBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
811
+ ],
812
+ "multi-gpu": [
813
+ "tests/models/plbart/test_modeling_plbart.py::PLBartModelTest::test_flash_attn_2_fp32_ln",
814
+ "tests/models/plbart/test_modeling_plbart.py::PLBartStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
815
+ ]
816
+ },
817
+ "qwen2": {
818
+ "single-gpu": [
819
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2ModelTest::test_flash_attn_2_fp32_ln",
820
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2IntegrationTest::test_model_450m_long_prompt"
821
+ ],
822
+ "multi-gpu": [
823
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2ModelTest::test_flash_attn_2_fp32_ln",
824
+ "tests/models/qwen2/test_modeling_qwen2.py::Qwen2IntegrationTest::test_model_450m_long_prompt"
825
+ ]
826
+ },
827
+ "qwen2_5_omni": {
828
+ "single-gpu": [
829
+ "tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py::Qwen2_5OmniThinkerForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
830
+ ],
831
+ "multi-gpu": [
832
+ "tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py::Qwen2_5OmniThinkerForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
833
+ ]
834
+ },
835
+ "qwen2_audio": {
836
+ "single-gpu": [
837
+ "tests/models/qwen2_audio/test_modeling_qwen2_audio.py::Qwen2AudioForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
838
+ ],
839
+ "multi-gpu": [
840
+ "tests/models/qwen2_audio/test_modeling_qwen2_audio.py::Qwen2AudioForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
841
+ ]
842
+ },
843
+ "qwen2_moe": {
844
+ "single-gpu": [
845
+ "tests/models/qwen2_moe/test_modeling_qwen2_moe.py::Qwen2MoeModelTest::test_flash_attn_2_fp32_ln"
846
+ ],
847
+ "multi-gpu": [
848
+ "tests/models/qwen2_moe/test_modeling_qwen2_moe.py::Qwen2MoeModelTest::test_flash_attn_2_fp32_ln"
849
+ ]
850
+ },
851
+ "qwen3": {
852
+ "single-gpu": [
853
+ "tests/models/qwen3/test_modeling_qwen3.py::Qwen3ModelTest::test_flash_attn_2_fp32_ln",
854
+ "tests/models/qwen3/test_modeling_qwen3.py::Qwen3IntegrationTest::test_model_600m_long_prompt"
855
+ ],
856
+ "multi-gpu": [
857
+ "tests/models/qwen3/test_modeling_qwen3.py::Qwen3ModelTest::test_flash_attn_2_fp32_ln"
858
+ ]
859
+ },
860
+ "qwen3_moe": {
861
+ "single-gpu": [
862
+ "tests/models/qwen3_moe/test_modeling_qwen3_moe.py::Qwen3MoeModelTest::test_flash_attn_2_fp32_ln"
863
+ ],
864
+ "multi-gpu": [
865
+ "tests/models/qwen3_moe/test_modeling_qwen3_moe.py::Qwen3MoeModelTest::test_flash_attn_2_fp32_ln"
866
+ ]
867
+ },
868
+ "qwen3_omni_moe": {
869
+ "single-gpu": [
870
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniMoeThinkerForConditionalGenerationModelTest::test_flash_attention_2_padding_matches_padding_free_with_position_ids",
871
+ "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",
872
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniMoeThinkerForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln",
873
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniModelIntegrationTest::test_small_model_integration_test_batch_flashatt2"
874
+ ],
875
+ "multi-gpu": [
876
+ "tests/models/qwen3_omni_moe/test_modeling_qwen3_omni_moe.py::Qwen3OmniMoeThinkerForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
877
+ ]
878
+ },
879
+ "qwen3_vl": {
880
+ "single-gpu": [
881
+ "tests/models/qwen3_vl/test_modeling_qwen3_vl.py::Qwen3VLModelTest::test_flash_attn_2_fp32_ln"
882
+ ],
883
+ "multi-gpu": [
884
+ "tests/models/qwen3_vl/test_modeling_qwen3_vl.py::Qwen3VLModelTest::test_flash_attn_2_fp32_ln"
885
+ ]
886
+ },
887
+ "qwen3_vl_moe": {
888
+ "single-gpu": [
889
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeModelTest::test_flash_attn_2_fp32_ln"
890
+ ],
891
+ "multi-gpu": [
892
+ "tests/models/qwen3_vl_moe/test_modeling_qwen3_vl_moe.py::Qwen3VLMoeModelTest::test_flash_attn_2_fp32_ln"
893
+ ]
894
+ },
895
+ "roberta": {
896
+ "single-gpu": [
897
+ "tests/models/roberta/test_modeling_roberta.py::RobertaModelTest::test_flash_attn_2_fp32_ln"
898
+ ],
899
+ "multi-gpu": [
900
+ "tests/models/roberta/test_modeling_roberta.py::RobertaModelTest::test_flash_attn_2_fp32_ln",
901
+ "tests/models/roberta/test_modeling_roberta.py::RobertaModelTest::test_flash_attn_2_inference_equivalence"
902
+ ]
903
+ },
904
+ "roberta_prelayernorm": {
905
+ "single-gpu": [
906
+ "tests/models/roberta_prelayernorm/test_modeling_roberta_prelayernorm.py::RobertaPreLayerNormModelTest::test_flash_attn_2_fp32_ln"
907
+ ],
908
+ "multi-gpu": [
909
+ "tests/models/roberta_prelayernorm/test_modeling_roberta_prelayernorm.py::RobertaPreLayerNormModelTest::test_flash_attn_2_fp32_ln"
910
+ ]
911
+ },
912
+ "seed_oss": {
913
+ "single-gpu": [
914
+ "tests/models/seed_oss/test_modeling_seed_oss.py::SeedOssModelTest::test_flash_attn_2_fp32_ln"
915
+ ],
916
+ "multi-gpu": [
917
+ "tests/models/seed_oss/test_modeling_seed_oss.py::SeedOssModelTest::test_flash_attn_2_fp32_ln"
918
+ ]
919
+ },
920
+ "smollm3": {
921
+ "single-gpu": [
922
+ "tests/models/smollm3/test_modeling_smollm3.py::SmolLM3ModelTest::test_flash_attn_2_fp32_ln"
923
+ ],
924
+ "multi-gpu": [
925
+ "tests/models/smollm3/test_modeling_smollm3.py::SmolLM3ModelTest::test_flash_attn_2_fp32_ln"
926
+ ]
927
+ },
928
+ "smolvlm": {
929
+ "single-gpu": [
930
+ "tests/models/smolvlm/test_modeling_smolvlm.py::SmolVLMForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
931
+ ],
932
+ "multi-gpu": [
933
+ "tests/models/smolvlm/test_modeling_smolvlm.py::SmolVLMForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
934
+ ]
935
+ },
936
+ "stablelm": {
937
+ "single-gpu": [
938
+ "tests/models/stablelm/test_modeling_stablelm.py::StableLmModelTest::test_flash_attn_2_fp32_ln",
939
+ "tests/models/stablelm/test_modeling_stablelm.py::StableLmModelIntegrationTest::test_model_3b_long_prompt"
940
+ ],
941
+ "multi-gpu": [
942
+ "tests/models/stablelm/test_modeling_stablelm.py::StableLmModelTest::test_flash_attn_2_fp32_ln"
943
+ ]
944
+ },
945
+ "starcoder2": {
946
+ "single-gpu": [
947
+ "tests/models/starcoder2/test_modeling_starcoder2.py::Starcoder2ModelTest::test_flash_attn_2_fp32_ln"
948
+ ],
949
+ "multi-gpu": [
950
+ "tests/models/starcoder2/test_modeling_starcoder2.py::Starcoder2ModelTest::test_flash_attn_2_fp32_ln"
951
+ ]
952
+ },
953
+ "t5gemma": {
954
+ "single-gpu": [
955
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_fp32_ln"
956
+ ],
957
+ "multi-gpu": [
958
+ "tests/models/t5gemma/test_modeling_t5gemma.py::T5GemmaModelTest::test_flash_attn_2_fp32_ln"
959
+ ]
960
+ },
961
+ "vaultgemma": {
962
+ "single-gpu": [
963
+ "tests/models/vaultgemma/test_modeling_vaultgemma.py::VaultGemmaModelTest::test_flash_attn_2_fp32_ln"
964
+ ],
965
+ "multi-gpu": [
966
+ "tests/models/vaultgemma/test_modeling_vaultgemma.py::VaultGemmaModelTest::test_flash_attn_2_fp32_ln"
967
+ ]
968
+ },
969
+ "video_llama_3": {
970
+ "single-gpu": [
971
+ "tests/models/video_llama_3/test_modeling_video_llama_3.py::VideoLlama3ModelTest::test_flash_attn_2_fp32_ln"
972
+ ],
973
+ "multi-gpu": [
974
+ "tests/models/video_llama_3/test_modeling_video_llama_3.py::VideoLlama3ModelTest::test_flash_attn_2_fp32_ln"
975
+ ]
976
+ },
977
+ "video_llava": {
978
+ "single-gpu": [
979
+ "tests/models/video_llava/test_modeling_video_llava.py::VideoLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
980
+ ],
981
+ "multi-gpu": [
982
+ "tests/models/video_llava/test_modeling_video_llava.py::VideoLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
983
+ ]
984
+ },
985
+ "vipllava": {
986
+ "single-gpu": [
987
+ "tests/models/vipllava/test_modeling_vipllava.py::VipLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
988
+ ],
989
+ "multi-gpu": [
990
+ "tests/models/vipllava/test_modeling_vipllava.py::VipLlavaForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
991
+ ]
992
+ },
993
+ "voxtral": {
994
+ "single-gpu": [
995
+ "tests/models/voxtral/test_modeling_voxtral.py::VoxtralForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
996
+ ],
997
+ "multi-gpu": [
998
+ "tests/models/voxtral/test_modeling_voxtral.py::VoxtralForConditionalGenerationModelTest::test_flash_attn_2_fp32_ln"
999
+ ]
1000
+ },
1001
+ "whisper": {
1002
+ "single-gpu": [
1003
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_flash_attn_2_fp32_ln",
1004
+ "tests/models/whisper/test_modeling_whisper.py::WhisperStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1005
+ ],
1006
+ "multi-gpu": [
1007
+ "tests/models/whisper/test_modeling_whisper.py::WhisperModelTest::test_flash_attn_2_fp32_ln",
1008
+ "tests/models/whisper/test_modeling_whisper.py::WhisperStandaloneDecoderModelTest::test_flash_attn_2_fp32_ln"
1009
+ ]
1010
+ },
1011
+ "xlm_roberta_xl": {
1012
+ "single-gpu": [
1013
+ "tests/models/xlm_roberta_xl/test_modeling_xlm_roberta_xl.py::XLMRobertaXLModelTest::test_flash_attn_2_fp32_ln"
1014
+ ],
1015
+ "multi-gpu": [
1016
+ "tests/models/xlm_roberta_xl/test_modeling_xlm_roberta_xl.py::XLMRobertaXLModelTest::test_flash_attn_2_fp32_ln"
1017
+ ]
1018
+ },
1019
+ "xmod": {
1020
+ "single-gpu": [
1021
+ "tests/models/xmod/test_modeling_xmod.py::XmodModelTest::test_flash_attn_2_fp32_ln"
1022
+ ],
1023
+ "multi-gpu": [
1024
+ "tests/models/xmod/test_modeling_xmod.py::XmodModelTest::test_flash_attn_2_fp32_ln"
1025
+ ]
1026
+ },
1027
+ "zamba2": {
1028
+ "single-gpu": [
1029
+ "tests/models/zamba2/test_modeling_zamba2.py::Zamba2ModelTest::test_flash_attn_2_fp32_ln"
1030
+ ],
1031
+ "multi-gpu": [
1032
+ "tests/models/zamba2/test_modeling_zamba2.py::Zamba2ModelTest::test_flash_attn_2_fp32_ln"
1033
+ ]
1034
+ }
1035
+ }