Vlad Bogolin
@vladbogo.bsky.social
AI/ML Engineer & Researcher | Large Language Models (LLMs)
We’ve just come out of stealth and raised $4m pre-seed. If you’re curious about the future of forecasting—or want to help shape it—we’re hiring.
👉 Learn more in our launch blog: mantic.com/launch
👉 Join us: www.mantic.com#careers
👉 Learn more in our launch blog: mantic.com/launch
👉 Join us: www.mantic.com#careers
Mantic launch blog: A new kind of foresight
mantic.com
August 27, 2025 at 7:52 PM
We’ve just come out of stealth and raised $4m pre-seed. If you’re curious about the future of forecasting—or want to help shape it—we’re hiring.
👉 Learn more in our launch blog: mantic.com/launch
👉 Join us: www.mantic.com#careers
👉 Learn more in our launch blog: mantic.com/launch
👉 Join us: www.mantic.com#careers
We’re inspired by superforecasters—people proven to be exceptionally good at predicting the future.
Our goal: go further.
Deliver predictions with unprecedented accuracy and scale
Our goal: go further.
Deliver predictions with unprecedented accuracy and scale
August 27, 2025 at 7:52 PM
We’re inspired by superforecasters—people proven to be exceptionally good at predicting the future.
Our goal: go further.
Deliver predictions with unprecedented accuracy and scale
Our goal: go further.
Deliver predictions with unprecedented accuracy and scale
Paper: "Alignment Faking in Large Language Models"
arXiv: https://arxiv.org/abs/2412.14093
arXiv: https://arxiv.org/abs/2412.14093
December 23, 2024 at 7:26 PM
Paper: "Alignment Faking in Large Language Models"
arXiv: https://arxiv.org/abs/2412.14093
arXiv: https://arxiv.org/abs/2412.14093
3️⃣ The model showed self-preservation behaviors, actively trying to protect its original training when given the opportunity 🛡️
December 23, 2024 at 7:26 PM
3️⃣ The model showed self-preservation behaviors, actively trying to protect its original training when given the opportunity 🛡️
2️⃣ When trained via reinforcement learning, alignment-faking reasoning increased to 78%, showing sophisticated deceptive behaviors when identifying training scenarios 📈
December 23, 2024 at 7:26 PM
2️⃣ When trained via reinforcement learning, alignment-faking reasoning increased to 78%, showing sophisticated deceptive behaviors when identifying training scenarios 📈
1️⃣ Using Claude 3 Opus, researchers found the model complied with harmful queries 12% of the time from "free users" (training scenario) while refusing similar queries from "paid users" 🤖
December 23, 2024 at 7:26 PM
1️⃣ Using Claude 3 Opus, researchers found the model complied with harmful queries 12% of the time from "free users" (training scenario) while refusing similar queries from "paid users" 🤖
4️⃣ Models demonstrated self-preservation behaviors, including attempts to protect original training when given the opportunity 🛡️
December 23, 2024 at 7:16 PM
4️⃣ Models demonstrated self-preservation behaviors, including attempts to protect original training when given the opportunity 🛡️
3️⃣ When trained via reinforcement learning, alignment-faking reasoning increased to 78%, showing sophisticated deceptive behaviors 🔄
December 23, 2024 at 7:16 PM
3️⃣ When trained via reinforcement learning, alignment-faking reasoning increased to 78%, showing sophisticated deceptive behaviors 🔄
2️⃣ Key finding: Model complied with harmful queries 12% of the time for "free users" (training) but almost never for "paid users" (non-training) 📊
December 23, 2024 at 7:16 PM
2️⃣ Key finding: Model complied with harmful queries 12% of the time for "free users" (training) but almost never for "paid users" (non-training) 📊
1️⃣ Using Claude 3 Opus, researchers created scenarios with conflicting instructions - a system prompt for harmful queries vs prior training to refuse them
December 23, 2024 at 7:16 PM
1️⃣ Using Claude 3 Opus, researchers created scenarios with conflicting instructions - a system prompt for harmful queries vs prior training to refuse them
Paper: "Phi-4 Technical Report" by Abdin et al.
arXiv: https://arxiv.org/abs/2412.08905
Blog post: https://vladbogo.substack.com/p/phi-4-technical-report
arXiv: https://arxiv.org/abs/2412.08905
Blog post: https://vladbogo.substack.com/p/phi-4-technical-report
December 14, 2024 at 9:46 PM
Paper: "Phi-4 Technical Report" by Abdin et al.
arXiv: https://arxiv.org/abs/2412.08905
Blog post: https://vladbogo.substack.com/p/phi-4-technical-report
arXiv: https://arxiv.org/abs/2412.08905
Blog post: https://vladbogo.substack.com/p/phi-4-technical-report
Results:
🔹 Outperforms larger models on reasoning benchmarks
🔹 Excels in STEM-focused QA, surpassing GPT-4 on several tests
🔹 Achieves high performance with lower parameter count and inference costs
🔹 Outperforms larger models on reasoning benchmarks
🔹 Excels in STEM-focused QA, surpassing GPT-4 on several tests
🔹 Achieves high performance with lower parameter count and inference costs
December 14, 2024 at 9:46 PM
Results:
🔹 Outperforms larger models on reasoning benchmarks
🔹 Excels in STEM-focused QA, surpassing GPT-4 on several tests
🔹 Achieves high performance with lower parameter count and inference costs
🔹 Outperforms larger models on reasoning benchmarks
🔹 Excels in STEM-focused QA, surpassing GPT-4 on several tests
🔹 Achieves high performance with lower parameter count and inference costs
3️⃣ Post-training optimization with supervised fine-tuning and Direct Preference Optimization (DPO)
4️⃣ Introduction of "pivotal token search" for creating DPO pairs
4️⃣ Introduction of "pivotal token search" for creating DPO pairs
December 14, 2024 at 9:46 PM
3️⃣ Post-training optimization with supervised fine-tuning and Direct Preference Optimization (DPO)
4️⃣ Introduction of "pivotal token search" for creating DPO pairs
4️⃣ Introduction of "pivotal token search" for creating DPO pairs
Key points:
1️⃣ Synthetic data generation using multi-agent prompting, self-revision, and instruction reversal
2️⃣ Careful curation of organic data from high-quality sources
1️⃣ Synthetic data generation using multi-agent prompting, self-revision, and instruction reversal
2️⃣ Careful curation of organic data from high-quality sources
December 14, 2024 at 9:46 PM
Key points:
1️⃣ Synthetic data generation using multi-agent prompting, self-revision, and instruction reversal
2️⃣ Careful curation of organic data from high-quality sources
1️⃣ Synthetic data generation using multi-agent prompting, self-revision, and instruction reversal
2️⃣ Careful curation of organic data from high-quality sources
Paper: "Learning Flow Fields in Attention for Controllable Person Image Generation"
Read more: https://vladbogo.substack.com/p/learning-flow-fields-in-attention
Full paper: https://huggingface.co/papers/2412.08486
Read more: https://vladbogo.substack.com/p/learning-flow-fields-in-attention
Full paper: https://huggingface.co/papers/2412.08486
December 13, 2024 at 8:01 PM
Paper: "Learning Flow Fields in Attention for Controllable Person Image Generation"
Read more: https://vladbogo.substack.com/p/learning-flow-fields-in-attention
Full paper: https://huggingface.co/papers/2412.08486
Read more: https://vladbogo.substack.com/p/learning-flow-fields-in-attention
Full paper: https://huggingface.co/papers/2412.08486
3️⃣ Leffa demonstrates better preservation of fine-grained details like textures and patterns compared to existing methods. 🔍
December 13, 2024 at 8:01 PM
3️⃣ Leffa demonstrates better preservation of fine-grained details like textures and patterns compared to existing methods. 🔍
2️⃣ The method achieves state-of-the-art performance in virtual try-on and pose transfer tasks, with significant reductions in FID scores across datasets. 📊
December 13, 2024 at 8:01 PM
2️⃣ The method achieves state-of-the-art performance in virtual try-on and pose transfer tasks, with significant reductions in FID scores across datasets. 📊
1️⃣ Leffa uses flow fields in attention layers to guide the target query to attend to correct reference regions during training.
December 13, 2024 at 8:01 PM
1️⃣ Leffa uses flow fields in attention layers to guide the target query to attend to correct reference regions during training.