Vladislav Moskvin

Azamat1k

Azamat1k159

AI & ML interests

None yet

Recent Activity

liked a model 5 days ago

google/gemma-3-27b-it

reacted to stefan-it's post with 👍 about 1 month ago

She arrived 😍 [Expect more models soon...]

reacted to ahmed-masry's post with 👍 about 2 months ago

Happy to announce AlignVLM 📏 – a novel approach to bridging vision and language latent spaces for multimodal understanding in Vision-Language Models (VLMs) 🌍📄🖼 🔗 Read the paper: https://huggingface.co/papers/2502.01341 🧐 What’s the challenge? Aligning visual features with language embeddings remains a major bottleneck in VLMs. Existing connectors such as Multi-layer perceptron (MLPs) often introduce noise that degrades performance. ❌ 🎯 Our Solution: ALIGN Connector We propose AlignVLM, a method that maps vision features into a weighted average of LLM text embeddings, ensuring they remain in a space that the LLM can effectively interpret. ✅ 🔬 How does it perform? We compared ALIGN against common connectors like MLPs, Perceiver Resampler, and Ovis trained under similar configurations. The results? ALIGN outperforms them all 🏆 on diverse document understanding tasks 📄. 📊 Meet the AlignVLM Model Family! We trained Llama 3.1 (1B, 3B, 8B) using our connector and benchmarked them against various models. The results: ✅ AlignVLM surpasses all Base VLMs trained under similar configurations. ✅ Our models also perform competitively against Instruct VLMs such as Qwen2-VL and InternVL-2.5 🚀. 🤔 What about robustness to noise? We injected Gaussian noise (μ=0, σ=3) into the vision encoder’s outputs before feeding them to the connector: ✅ ALIGN Connector: Minimal drop (↓1.67%) – proving its high robustness! ❌ MLP Connector: Severe degradation (↓25.54%) – struggling with noisy inputs. Code & model weights coming soon! Stay tuned! 🔥

View all activity

Organizations

None yet

Azamat1k's activity

liked a model 5 days ago

google/gemma-3-27b-it

Image-Text-to-Text • Updated 15 days ago • 977k • • 1.11k

liked a dataset about 2 months ago

iluvvatar/NEREL

Updated Mar 30, 2023 • 70 • 6

liked a model about 2 months ago

inkoziev/chargpt-96M

Text Generation • Updated Sep 2, 2023 • 205 • 3

liked 3 models 2 months ago

liked 2 models 3 months ago

MiniMaxAI/MiniMax-VL-01

Image-Text-to-Text • Updated 19 days ago • 174 • 251

openbmb/MiniCPM-o-2_6

Any-to-Any • Updated 9 days ago • 854k • 1.08k

liked a model 4 months ago

msu-rcc-lair/RuadaptQwen2.5-32B-instruct-GGUF

Text Generation • Updated Dec 31, 2024 • 650 • 18

liked 2 Spaces 4 months ago

RuadaptQwen2.5

💬

Communicate with a helpful AI assistant

5.31k

MTEB Leaderboard

🥇

Embedding Leaderboard

liked a Space 5 months ago

414

LLM Model VRAM Calculator

📈

Calculate VRAM requirements for running large language models

liked 2 models 5 months ago

ai-forever/FRED-T5-1.7B

Text2Text Generation • Updated Dec 5, 2023 • 1.22k • • 76

Vikhrmodels/Vikhr-Llama-3.2-1B-Instruct

Text Generation • Updated Sep 27, 2024 • 3.85k • 40

liked a Space 5 months ago

Russian LLM Leaderboard

🥇

View and submit LLM benchmark evaluations

liked a model 10 months ago

urchade/gliner_multi-v2.1

Token Classification • Updated Apr 10, 2024 • 12k • 118

liked 3 models 11 months ago

nvidia/Llama3-ChatQA-1.5-8B

Text Generation • Updated May 24, 2024 • 5.83k • 555

IlyaGusev/saiga_llama3_8b_gguf

Text Generation • Updated Jul 4, 2024 • 1.27k • 79

IlyaGusev/saiga_llama3_8b

Text Generation • Updated Jul 4, 2024 • 33.2k • 121

liked a model 12 months ago

nateraw/musicgen-songstarter-v0.2

Text-to-Audio • Updated May 3, 2024 • 9.71k • 153