 Anything
9.0K
32.0M
LLaVA-OV 1.5 Instruct
1.6M
4.1B
PixMo Cap QA
25.3K
120.3M
LLaVA-OV 2 (OSD)
118.3K
259.3M
PubMedVision
1.3M
1.9B
LLaVA-OV 2 (RoboRefer)
37.2K
102.1M
RSRCC
15.7K
36.7M
LNQA (recap)
45.9K
156.7M
SenseNova-SI-8M
1.0M
8.0B
MapTrace
292.0K
2.5B
SenseNova-SI-8M (CoT)
37.4K
3.0B
Memotion (English recap)
887
1.5M
ShizhenGPT IT
10.1K
32.1M
MMEvol (judged)
14.0K
32.7M
SpatialSense (gold)
1.0K
974.2K
Molmo2-MultiImageQA
3.8K
30.1M
VDR Cooking Recipes
1.8K
12.1M
Molmo2-SynMultiImageQA
15.1K
349.1M
Appendix C Audio Datasets by Training Stage
Tables 0, 0, 0, 0, and 0 give the audio pool of each training stage. Appendix F and Table 8 give the corresponding per-run mixtures for the long-context extension and long-context SFT.
Table C.1: Audio pool of Stage 1.
Dataset
Samples
Tokens
Dataset
Samples
Tokens
AudioSet (bal./unbal.)
1.7M
700.3M
Suno
235.4K
1.4B
Common Voice 24
2.0M
427.0M
People’s Speech (unsup.)
15.5M
51.2B
Gemeinderat Zurich
72.1K
172.7M
VoxPopuli
31.9M
56.7B
MTG-Jamendo
19.5K
107.6M
Table C.2: Audio pool of Stage 2 and the cooldown.
Dataset
Samples
Tokens
Dataset
Samples
Tokens
AISHELL-1/3/4
226.7K
48.8M
MediaSpeech
10.0K
6.3M
Aozora Hurigana
2.6M
703.6M
MultiMed
34.1K
15.2M
CC-Podcasts
238.5K
437.4M
OmniLingual ASR
183.9K
373.9M
Common Voice 24
11.7M
2.6B
ParlaMent Parla
241.7K
115.8M
Coral V3
295.2K
80.6M
ParlaSpeech-RS
159.7K
117.7M
EuroSpeech
2.1M
2.1B
People’s Speech (sup.)
2.1M
1.4B
F1 Team Radio
14.3K
6.8M
Russian LibriSpeech
31.3K
14.6M
Farsi ASR
329.7K
78.7M
SPC-R
66.1K
59.3M
Ghana English Speech
405.9K
96.6M
SeamlessAlign
6.0M
2.4B
GigaSpeech
7.8M
1.6B
SloPalSpeech
243.3K
450.1M
GigaSpeech 2
19.0M
3.9B
ToneWebinars
285.2K
325.3M
Granary (YODAS/YTC)
74.1M
41.9B
Tuda-De
50.8K
19.2M
Infore2 Audiobooks
265.8K
68.3M
ViMedCSS
7.3K
3.8M
Kathbath
776.0K
240.3M
VietSpeech
1.0M
183.4M
Kazakh Speech
230.7K
75.6M
VoxPopuli
22.2M
28.1B
LegCo Speech
3.5M
1.4B
WenetSpeech
5.9M
761.5M
LibriHeavy
4.5M
3.0B
Zeroth Korean
21.8K
8.0M
Localized Narratives Audio
827.6K
707.8M
Zoengjyutgaai
47.0K
14.4M
MLS
580.0K
380.1M
Table C.3: Audio pool of the long-context extension.
Dataset
Samples
Tokens
Dataset
Samples
Tokens
AISHELL-1/3/4
39
405.6K
ParlaSpeech-RS
286
2.9M
EuroSpeech
16.6K
206.2M
People’s Speech (sup.)
66.8K
1.8B
Granary (YODAS/YTC)
266.2K
3.3B
SPC-R
1.2K
16.0M
LegCo Speech
53.8K
1.7B
VoxPopuli
509.7K
30.2B
LibriHeavy
146.4K
4.1B
WenetSpeech
42.6K
841.1M
MLS
3.5K
556.9M
Zoengjyutgaai
838
16.3M
Dataset
Samples
Tokens
HeySQuAD human
49.6K
23.5M
HeySQuAD machine
98.2K
36.2M
TCM Shizhen Speech
58.4K
59.5M
TeleAntiFraud
33.0K
124.0M
Vocalized SFT
635.1K
725.6M
VoiceAssistant-400K
242.2K
144.3M
Table C.4: Audio pool of SFT.
Dataset
Samples
Tokens
AudioMCQ GeminiCoT
10.5K
6.4M
HeySQuAD human
49.6K
23.5M
HeySQuAD machine
98.2K
36.2M
Marco Longspeech
204.9K
5.5B
TCM Shizhen Speech
58.4K
59.5M
TeleAntiFraud
34.0K
133.0M
Vocalized SFT
635.3K
727.2M
VoiceAssistant-400K
242.2K
144.3M
Table C.5: Audio pool of long-context SFT.
Appendix D Image Dataset Exploration Detailed Breakdown
Our large-scale multimodal image collections comprise a diverse range of visual media, including photographs, illustrations, screenshots, documents, diagrams, and composite graphics, with correspondingly varying content. While visual benchmarks are usually balanced across different image types and different task types, our vision data is not guaranteed to be balanced across image types, and might therefore cover some benchmarks poorly. Beyond the coarse classification of Figure 3, quantifying our coverage of different image types thus requires a rich representation taxonomy. In turn, this could lead to a filtering scheme that balances the vision data in appropriate ways.
Taxonomy. We therefore design a taxonomy consisting of 61 fine-grained labels organized along two orthogonal dimensions: graphics and photographs. We also organize these fine-grained labels into five coarse-grained labels for graphics and seven coarse-grained labels for photographs f