Sotwe logo Sotwe logo

Top Tweets for #OpenBMB

OpenBMB's profile image
OpenBMB Verified account @OpenBMB
OpenBMB Open Source Week is coming. ๐Ÿš€ ๐Ÿ“… May 22โ€“28 No secrets. Just full-stack Edge AI. We are opening our entire stack: โš™๏ธ Infrastructure ๐Ÿง  Models ๐Ÿ‹๏ธ Training ๐Ÿ“ฑ Applications ๐Ÿ“Š Datasets Everything open on GitHub & Hugging Face. See you there. #OpenBMB #EdgeAI #OpenSource
OpenBMB's tweet photo. OpenBMB Open Source Week is coming. ๐Ÿš€ ๐Ÿ“… May 22โ€“28 No secrets. Just full-stack Edge AI. We are opening our entire stack: โš™๏ธ Infrastructure ๐Ÿง  Models ๐Ÿ‹๏ธ Training ๐Ÿ“ฑ Applications ๐Ÿ“Š Datasets Everything open on GitHub & Hugging Face. See you there. #OpenBMB #EdgeAI #OpenSource
2
23
5
2
2K
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
๐Ÿš€ VoxCPM 2 is trending on @HuggingFace๏ผ Tokenizer-Free #TTS for multilingual speech generationโ€” now getting real traction ๐ŸŽ™๏ธ Appreciate all the love from the community ๐Ÿฅฐ ๐Ÿ”— Try it: https://t.co/vhnf1fY6pT https://t.co/7gdD0R0PIc #TTS #AI #OpenSource #OpenBMB
OpenBMB's tweet photo. ๐Ÿš€ VoxCPM 2 is trending on @HuggingFace๏ผ Tokenizer-Free #TTS for multilingual speech generationโ€” now getting real traction ๐ŸŽ™๏ธ Appreciate all the love from the community ๐Ÿฅฐ ๐Ÿ”— Try it: https://t.co/vhnf1fY6pT https://t.co/7gdD0R0PIc #TTS #AI #OpenSource #OpenBMB https://t.co/R88Wrdjq31
6
57
11
17
4K
Sarah_GreenOk's profile image
Sarah Green Verified account @Sarah_GreenOk
@OpenBMB @TsinghuaNLP Hey #OpenBMB, love the team youโ€™re building! ๐Ÿ’–
0
0
0
0
28
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
When #LLMs surpass human experts in math and coding, who provides the ground truth? How can we train models when human supervision becomes the bottleneck? ๐Ÿค” Today, we dive into Unsupervised RLVRโ€”a comprehensive new study by @TsinghuaNLP (an #OpenBMB member)alongside researchers from Shanghai AI Lab, XJTU, and UIUC. This paper systematically answers exactly how far unsupervised verifiable rewards can scale LLM training. ๐Ÿค— Paper: https://t.co/LXiVlllD62 ๐Ÿ“„ arXiv: https://t.co/z2zlFcJ4fm ๐Ÿ’ป Code: https://t.co/KhvWeG8zMk Why it matters: 1๏ธโƒฃ The "Sharpening" Illusion: We mathematically prove that methods relying on intrinsic rewards (like model confidence or consistency) don't actually learn new knowledge. They merely "sharpen" the model's initial preference, converging into a deterministic policy. ๐Ÿช’ 2๏ธโƒฃ The "Rise and Fall" Trap: Because of this sharpening effect, intrinsic RLVR inevitably faces a collapse. It improves initially when the model's intuition is right, but quickly leads to "reward hacking" where proxy rewards rise while true accuracy plummets. ๐Ÿ“‰ 3๏ธโƒฃ A New Metric for RL Potential: We introduce "Model Collapse Steps" (the steps it takes for reward accuracy to drop below 1%). This surprisingly stable metric acts as a highly efficient predictor of a model's prior capacity and trainability for RL! โฑ๏ธ 4๏ธโƒฃ The Path Forward: While intrinsic rewards hit a wall, our exploration of extrinsic rewards (like self-verification) shows true promise in breaking these limitations, paving the way for sustainable LLM self-evolution. ๐Ÿš€ To scale beyond human limits, we must rethink what rewards we trust. Read the full paper to see our theoretical framework and extensive veRL experiments! #AI #ReinforcementLearning #RLVR #MachineLearning
OpenBMB's tweet photo. When #LLMs surpass human experts in math and coding, who provides the ground truth? How can we train models when human supervision becomes the bottleneck? ๐Ÿค” Today, we dive into Unsupervised RLVRโ€”a comprehensive new study by @TsinghuaNLP (an #OpenBMB member)alongside researchers from Shanghai AI Lab, XJTU, and UIUC. This paper systematically answers exactly how far unsupervised verifiable rewards can scale LLM training. ๐Ÿค— Paper: https://t.co/LXiVlllD62 ๐Ÿ“„ arXiv: https://t.co/z2zlFcJ4fm ๐Ÿ’ป Code: https://t.co/KhvWeG8zMk Why it matters: 1๏ธโƒฃ The "Sharpening" Illusion: We mathematically prove that methods relying on intrinsic rewards (like model confidence or consistency) don't actually learn new knowledge. They merely "sharpen" the model's initial preference, converging into a deterministic policy. ๐Ÿช’ 2๏ธโƒฃ The "Rise and Fall" Trap: Because of this sharpening effect, intrinsic RLVR inevitably faces a collapse. It improves initially when the model's intuition is right, but quickly leads to "reward hacking" where proxy rewards rise while true accuracy plummets. ๐Ÿ“‰ 3๏ธโƒฃ A New Metric for RL Potential: We introduce "Model Collapse Steps" (the steps it takes for reward accuracy to drop below 1%). This surprisingly stable metric acts as a highly efficient predictor of a model's prior capacity and trainability for RL! โฑ๏ธ 4๏ธโƒฃ The Path Forward: While intrinsic rewards hit a wall, our exploration of extrinsic rewards (like self-verification) shows true promise in breaking these limitations, paving the way for sustainable LLM self-evolution. ๐Ÿš€ To scale beyond human limits, we must rethink what rewards we trust. Read the full paper to see our theoretical framework and extensive veRL experiments! #AI #ReinforcementLearning #RLVR #MachineLearning
0
22
5
11
2K
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
When #LLMs surpass human experts in math and coding, who provides the ground truth? How can we train models when human supervision becomes the bottleneck? ๐Ÿค” Today, we dive into Unsupervised RLVRโ€”a comprehensive new study by @TsinghuaNLP (an #OpenBMB member) alongside researchers from Shanghai AI Lab, XJTU, and UIUC. This paper systematically answers exactly how far unsupervised verifiable rewards can scale LLM training. ๐Ÿค— Paper: https://t.co/LXiVllmaVA ๐Ÿ“„ arXiv: https://t.co/z2zlFcJC4U ๐Ÿ’ป Code: https://t.co/KhvWeG97BS Why it matters: 1๏ธโƒฃ The "Sharpening" Illusion: We mathematically prove that methods relying on intrinsic rewards (like model confidence or consistency) don't actually learn new knowledge. They merely "sharpen" the model's initial preference, converging into a deterministic policy. ๐Ÿช’ 2๏ธโƒฃ The "Rise and Fall" Trap: Because of this sharpening effect, intrinsic RLVR inevitably faces a collapse. It improves initially when the model's intuition is right, but quickly leads to "reward hacking" where proxy rewards rise while true accuracy plummets. ๐Ÿ“‰ 3๏ธโƒฃ A New Metric for RL Potential: We introduce "Model Collapse Steps" (the steps it takes for reward accuracy to drop below 1%). This surprisingly stable metric acts as a highly efficient predictor of a model's prior capacity and trainability for RL! โฑ๏ธ 4๏ธโƒฃ The Path Forward: While intrinsic rewards hit a wall, our exploration of extrinsic rewards (like self-verification) shows true promise in breaking these limitations, paving the way for sustainable LLM self-evolution. ๐Ÿš€ To scale beyond human limits, we must rethink what rewards we trust. Read the full paper to see our theoretical framework and extensive veRL experiments! #AI #THUNLP #OpenBMB #LLM #ReinforcementLearning #RLVR #MachineLearning
OpenBMB's tweet photo. When #LLMs surpass human experts in math and coding, who provides the ground truth? How can we train models when human supervision becomes the bottleneck? ๐Ÿค” Today, we dive into Unsupervised RLVRโ€”a comprehensive new study by @TsinghuaNLP (an #OpenBMB member) alongside researchers from Shanghai AI Lab, XJTU, and UIUC. This paper systematically answers exactly how far unsupervised verifiable rewards can scale LLM training. ๐Ÿค— Paper: https://t.co/LXiVllmaVA ๐Ÿ“„ arXiv: https://t.co/z2zlFcJC4U ๐Ÿ’ป Code: https://t.co/KhvWeG97BS Why it matters: 1๏ธโƒฃ The "Sharpening" Illusion: We mathematically prove that methods relying on intrinsic rewards (like model confidence or consistency) don't actually learn new knowledge. They merely "sharpen" the model's initial preference, converging into a deterministic policy. ๐Ÿช’ 2๏ธโƒฃ The "Rise and Fall" Trap: Because of this sharpening effect, intrinsic RLVR inevitably faces a collapse. It improves initially when the model's intuition is right, but quickly leads to "reward hacking" where proxy rewards rise while true accuracy plummets. ๐Ÿ“‰ 3๏ธโƒฃ A New Metric for RL Potential: We introduce "Model Collapse Steps" (the steps it takes for reward accuracy to drop below 1%). This surprisingly stable metric acts as a highly efficient predictor of a model's prior capacity and trainability for RL! โฑ๏ธ 4๏ธโƒฃ The Path Forward: While intrinsic rewards hit a wall, our exploration of extrinsic rewards (like self-verification) shows true promise in breaking these limitations, paving the way for sustainable LLM self-evolution. ๐Ÿš€ To scale beyond human limits, we must rethink what rewards we trust. Read the full paper to see our theoretical framework and extensive veRL experiments! #AI #THUNLP #OpenBMB #LLM #ReinforcementLearning #RLVR #MachineLearning
1
15
1
7
1K
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Multi-Agent Systems (MAS) typically focus on who speaks (role assignment), but often overlook how they effectively communicate. Structural design isn't enough for strategic interaction. ๐Ÿค Today, we present LinguaGameโ€”new research from @TsinghuaNLP (#OpenBMB member), @UCBerkeley, and @PKU1898: A linguistically grounded, game-theoretic paradigm that treats dialogue as a strategic signaling game over intents and strategies. ๐Ÿค— Paper: https://t.co/qrpnBHXuRf ๐Ÿ“„ arXiv: https://t.co/p2gzhZCPn3 Why it matters: 1๏ธโƒฃLinguistics Meets Game Theory: Unlike rigid task-specific games, LinguaGame models dialogue as a Signaling Game grounded in Speech Act Theory. Agents don't just generate text; they explicitly infer what others want (Intents) and how they plan to achieve it (Strategies), leading to deeper mutual understanding. ๐Ÿง  2๏ธโƒฃTraining-Free Efficiency: No expensive model retraining required! LinguaGame introduces an inference-time equilibrium approximation algorithm. It acts as a plug-and-play layer that optimizes agent policies on the fly, making it compatible with existing LLM-based systems instantly. โšก๏ธ 3๏ธโƒฃSmarter, Strategic Agents: Tested in complex simulated Courtrooms and Debates, LinguaGame significantly boosts communication efficiency. Human expert evaluation shows agents produce clearer, more concise, and logically sound arguments compared to standard MAS architectures. โš–๏ธ๐Ÿ—ฃ๏ธ LinguaGame redefines agent interaction, moving from simple role-play to deep strategic reasoning for high-quality dialogue generation. #AI #THUNLP #LLM #MultiAgent #NLP #MachineLearning
OpenBMB's tweet photo. Multi-Agent Systems (MAS) typically focus on who speaks (role assignment), but often overlook how they effectively communicate. Structural design isn't enough for strategic interaction. ๐Ÿค Today, we present LinguaGameโ€”new research from @TsinghuaNLP (#OpenBMB member), @UCBerkeley, and @PKU1898: A linguistically grounded, game-theoretic paradigm that treats dialogue as a strategic signaling game over intents and strategies. ๐Ÿค— Paper: https://t.co/qrpnBHXuRf ๐Ÿ“„ arXiv: https://t.co/p2gzhZCPn3 Why it matters: 1๏ธโƒฃLinguistics Meets Game Theory: Unlike rigid task-specific games, LinguaGame models dialogue as a Signaling Game grounded in Speech Act Theory. Agents don't just generate text; they explicitly infer what others want (Intents) and how they plan to achieve it (Strategies), leading to deeper mutual understanding. ๐Ÿง  2๏ธโƒฃTraining-Free Efficiency: No expensive model retraining required! LinguaGame introduces an inference-time equilibrium approximation algorithm. It acts as a plug-and-play layer that optimizes agent policies on the fly, making it compatible with existing LLM-based systems instantly. โšก๏ธ 3๏ธโƒฃSmarter, Strategic Agents: Tested in complex simulated Courtrooms and Debates, LinguaGame significantly boosts communication efficiency. Human expert evaluation shows agents produce clearer, more concise, and logically sound arguments compared to standard MAS architectures. โš–๏ธ๐Ÿ—ฃ๏ธ LinguaGame redefines agent interaction, moving from simple role-play to deep strategic reasoning for high-quality dialogue generation. #AI #THUNLP #LLM #MultiAgent #NLP #MachineLearning
3
27
4
10
1K
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
๐Ÿš€AgentCPM-Report is leveling up #DeepResearch! Jointly developed by @TsinghuaNLP, RUC, @ModelBest2022, and #OpenBMB, this 8B SOTA agent delivers #Gemini-2.5-Pro level reports directly on your local machine. No cloud, no data leaksโ€”just pure, private intelligence. ๐Ÿ›ก๏ธ We want to hear from our fellow #researchers & #PhD students! ๐Ÿ‘‚๐ŸปWhat matters MOST to you in a local AI research assistant? (Tell us why in the comments! ๐Ÿ‘‡) If you find our work helpful, please support us with a โค๏ธ on @huggingface ! It helps the project reach more builders. ๐Ÿ‘‰๐Ÿปhttps://t.co/6zzxom5h5y #OpenSource #LLM #AIResearch #AgentCPM #DeepResearch #MachineLearning
4
42
10
32
14K
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
๐Ÿ’ฅ10K LIKES UNLOCKED! MiniCPM's roaring from 0.5B params to pure SOTA magic-thanks to YOU, MVPs. Let's keep the fire alive! ๐Ÿ”ฅ #MiniCPM #OpenBMB
OpenBMB's tweet photo. ๐Ÿ’ฅ10K LIKES UNLOCKED! MiniCPM's roaring from 0.5B params to pure SOTA magic-thanks to YOU, MVPs. Let's keep the fire alive! ๐Ÿ”ฅ #MiniCPM #OpenBMB https://t.co/9pMf7xtbZV
0
9
0
0
495
Rixhabh__'s profile image
Rishabh Verified account @Rixhabh__
Why does realistic voice matter? Because emotion lives in tone, not just text. VoxCPM brings AI to life: โœ… Natural prosody โœ… Humanlike flow โœ… Zero-shot voice cloning Try it free โ†’ https://t.co/hEQuLCCL7a #MiniCPM #TTS #OpenBMB
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
14
21
0
2
2K
Jubayer7792821's profile image
Jubayer Verified account @Jubayer7792821
๐Ÿšจ Next-level AI speech is here! Meet VoxCPM โ†’ built on MiniCPM-4 0.5B, trained on 1.8M+ hours. โœ”๏ธ Tokenizer-free TTS โœ”๏ธ Realistic voices โœ”๏ธ Zero-shot cloning ๐ŸŽ™๏ธ Try it ๐Ÿ‘‰ https://t.co/E5suX5l6tx #OpenBMB #AI #VoiceTech
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
1
4
1
1
258
Damn_coder's profile image
D-Coder Verified account @Damn_coder
Meet VoxCPM โ€” the tokenizer-free TTS model changing the game. Trained on 1.8M hours, powered by MiniCPM-4 0.5B, and built for: โœ… Hyper-realistic speech โœ… Zero-shot voice cloning โœ… Natural prosody & rhythm Try it here ๐Ÿ‘‰ https://t.co/xWcZOJadQQ #OpenBMB #MiniCPM #AI #TTS
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
3
8
0
1
965
james_bonds0's profile image
James Bond Verified account @james_bonds0
๐Ÿ† New SOTA in speech synthesis โ†’ VoxCPM. Built on MiniCPM-4 0.5B, trained on 1.8M+ hours. โœ… Realistic speech โœ… Zero-shot cloning โœ… Natural prosody ๐ŸŽง Try it on Hugging Face ๐Ÿ‘‰ https://t.co/g3bZ9S4xR0 #TTS #AI #OpenBMB https://t.co/iVkwiLGSwh
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
38
41
9
4
3K
code_everett's profile image
Lucas Everett Verified account @code_everett
๐Ÿ”ฅ VoxCPM = TTS Redefined Zero-shot cloning + lifelike prosody. Built on MiniCPM-4 0.5B, trained on 1.8M+ hours. ๐Ÿ‘‰https://t.co/Bp4e6btYjZ #AI #TTS #OpenBMB
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
0
6
0
0
36K
heyemilyx's profile image
Emily Watson | AI Tools & Tech News Verified account @heyemilyx
๐Ÿ”ฅ Meet VoxCPM: Our game-changing, tokenizer-free TTS model. Powered by MiniCPM-4 0.5B, it delivers hyper-realistic speech, zero-shot voice cloning, and natural prosody. Trained over 1.8M hours, hitting SOTA. Try it on huggingface ๐Ÿ‘‰ https://t.co/4pjZAAgFsj #OpenBMB #MiniCPM #AI #TTS
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
4
3
1
1
327
Ubermenscchh's profile image
Hasan Verified account @Ubermenscchh
VoxCPM is rewriting the TTS playbook ๐Ÿ“ Diffusion autoregressive architecture + semantic-acoustic decoupling = speech synthesis that sounds impossibly real Zero-shot voice cloning works like actual magic โœจ Try it here : https://t.co/Tewq5nUkSR #OpenBMB #MiniCPM #AIVoice
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
4
6
0
1
563
AIwithArsalan's profile image
Arsalan Verified account @AIwithArsalan
Check out VoxCPM, a new tokenizer-free TTS model built on MiniCPM-4 0.5B. - Creates lifelike speech - Clones voices from a short clip - Maintains natural flow Trained on 1.8M hours for next-level performance. Try it ๐Ÿ‘‰ https://t.co/FFEb4i6AhR #OpenBMB #MiniCPM #AI #TTS
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
9
16
0
2
1K
hasantoxr's profile image
Hasan Toor Verified account @hasantoxr
Meet VoxCPM: This is game-changing, tokenizer-free TTS model. Powered by MiniCPM-4 0.5B, it delivers hyper-realistic speech, zero-shot voice cloning, and natural prosody. Trained over 1.8M hours, hitting SOTA. Try it on huggingface https://t.co/WRgmFNxtNF #OpenBMB #MiniCPM #AI #TTS
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
0
2
0
5
4K
Ronycoder's profile image
Rony Verified account @Ronycoder
๐Ÿšจ Meet VoxCPM โ€” the next-gen, tokenizer-free TTS model. Powered by MiniCPM-4 0.5B, it delivers ultra-realistic speech, zero-shot voice cloning, and natural prosody. Trained on 1.8M+ hours for SOTA results. ๐Ÿ‘‰ Try it now: https://t.co/jCx5RpQKsp #OpenBMB #MiniCPM #AI #TTS
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
8
18
4
2
13K
HeyAmit_'s profile image
Amit Verified account @HeyAmit_
๐Ÿ”ฅ Meet VoxCPM: Our game-changing, tokenizer-free TTS model. Powered by MiniCPM-4 0.5B, it delivers hyper-realistic speech, zero-shot voice cloning, and natural prosody. Trained over 1.8M hours, hitting SOTA. Try it on huggingface ๐Ÿ‘‰ https://t.co/lzE1SepEiG #OpenBMB #MiniCPM #AI #TTS
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
Why does a realistic voice matter? ๐Ÿค”๐Ÿค” The same robot that feels creepy can transform into a trusted companion just by having a human-like voice. Think about the movie, Her. ๐Ÿ”ฅVoxCPM is a new paradigm: continuous, context-aware, and incredibly lifelike. โœ… Small size: 0.5B only โœ… Zero-Shot Voice Cloning โœ… Context-Aware, Expressive Speech Generation Try it ๐Ÿ‘‰ ๐Ÿ”—Huggingface | https://t.co/lZ6HTV2175 ๐Ÿ”—Github | https://t.co/aNXpfgrRt9
32
155
25
93
128K
13
18
1
2
2K
OpenBMB's profile image
OpenBMB Verified account @OpenBMB
โค๏ธ Thanks to @mervenoyann & @huggingface , MiniCPM-V 4.5 is officially live on Hugging Face Spaces. Come check it out๏ผ https://t.co/LyMRbBaekW Get ready for the future of multimodal AI ๐Ÿ‘‰ Huggingface๏ฝœhttps://t.co/JW7zPXOCqh #AI #MiniCM #GPT #Gemini #OpenBMB #ArtificialIntelligence #MachineLearning
2
30
2
8
3K

Last Seen Hashtags on Sotwe

prankcowokstraight
Seen from Indonesia
alanyagay
Seen from Turkey
ipcam
Seen from Canada
teenageer()filter:media
Seen from Brazil
ometv #ometv ometv
Seen from Turkey
groping #whore
Seen from Canada
ๅญๅฎซ
omegle filter:native_video since:2026-06-6
Seen from Greece
dilecehin
Seen from Indonesia

Trends for you

1
Everton
Under 10K tweets
2
Kimi
Under 10K tweets
3
Michigan
Under 10K tweets
4
Andy Williams
Under 10K tweets
5
Rashford
Under 10K tweets
6
Havertz
Under 10K tweets
7
Morgan Rogers
Under 10K tweets
8
Fermin
Under 10K tweets
9
Good Sunday
Under 10K tweets
10
Lammens
Under 10K tweets

Most Popular Users

elonmusk's profile image
1
Elon Musk Verified account
@elonmusk
241.6M followers
barackobama's profile image
2
Barack Obama Verified account
@barackobama
119M followers
cristiano's profile image
3
Cristiano Ronaldo Verified account
@cristiano
114.1M followers
realdonaldtrump's profile image
4
Donald J. Trump Verified account
@realdonaldtrump
111.8M followers
narendramodi's profile image
5
Narendra Modi Verified account
@narendramodi
107.2M followers
rihanna's profile image
6
Rihanna Verified account
@rihanna
98.7M followers
nasa's profile image
7
NASA Verified account
@nasa
92.4M followers
justinbieber's profile image
8
Justin Bieber Verified account
@justinbieber
91.8M followers
katyperry's profile image
9
KATY PERRY Verified account
@katyperry
89.8M followers
taylorswift13's profile image
10
Taylor Swift Verified account
@taylorswift13
83.8M followers
ladygaga's profile image
11
Lady Gaga Verified account
@ladygaga
75.3M followers
imvkohli's profile image
12
Virat Kohli Verified account
@imvkohli
73.1M followers
kimkardashian's profile image
13
Kim Kardashian Verified account
@kimkardashian
70.8M followers
youtube's profile image
14
YouTube Verified account
@youtube
68.8M followers
neymarjr's profile image
15
Neymar Jr Verified account
@neymarjr
66.1M followers
billgates's profile image
16
Bill Gates Verified account
@billgates
65M followers
selenagomez's profile image
17
Selena Gomez Verified account
@selenagomez
62.9M followers
theellenshow's profile image
18
The Ellen Show
@theellenshow
62.3M followers
cnn's profile image
19
CNN Verified account
@cnn
61.8M followers
x's profile image
20
X Verified account
@x
60.7M followers