Check out the Blogs
Click here!Hey there! Welcome to "My-Personal-Model-Picks" – basically my digital notebook for cool AI models I've stumbled upon.
So here's the deal: I kept losing track of these awesome, compact AI models that can run on your own machine (hello, privacy!) and are surprisingly versatile. Fed up with forgetting, I decided to dump them all here.
What you'll find:
This isn't some polished showcase – it's more like my messy desk of AI treasures. If you're into AI that's practical, privacy-conscious, and maybe a bit inspired by how our brains or rather nature work, you might dig this collection.
Whether you're a fellow student, a curious dev, or just someone who likes efficient tech, feel free to poke around. I'm constantly updating this as I find new gems or when procrastination hits hard during finals week.
No pressure to stick around, but if you're into this kind of stuff, you might find something neat. Who knows, maybe you'll discover your next project inspiration or a solution to that tricky AI problem you've been wrestling with.
Happy exploring, and may your models be ever accurate and your privacy intact! 🤖🔒✨
Model | Link | Type | Size | Disk Space | Immediate Use Cases |
---|---|---|---|---|---|
NuExtract-Tiny | HuggingFace (a not so tiny, 3.8B version) | Text-to-StructData | 494M params | 988Mb | Maybe, extract structured data from text documents with multilingual & long-form support |
OuteTTS-350M | HuggingFace | Text-to-Speech | 362M params | 724Mb | Maybe, pure LLM-based text-to-speech with multiple voices |
Jina-Embeddings-V2-Base-Code | HuggingFace | Text-Embeddings | 161M params | 322Mb | Maybe, code-specific embeddings with 8k sequence length support |
Mini-Omni | HuggingFace | Speech-to-Text, Speech+Text-to-Text | 0.5B params | 2.7Gb | Maybe, real-time transcription and translation for multilingual podcasts or videos |
MusicGen-Small | HuggingFace | Text-to-Music | 300M params | 2.36Gb | Maybe, music generation based on the themes of certain text |
RMBG | HuggingFace | Background-removal | * | 200Mb | Maybe, a step in automated video-editing |
Whisper | HuggingFace | Speech-to-Text | 74M | 290Mb | Maybe, another form of input beyond text |
SmolLM-360M | HuggingFace | Text-Gen | 360M | 724Mb | Maybe, simple instruction following |
NuExtract-Tiny | HuggingFace | Text-to-StructData | 0.5B | 1.86Gb | Fast structured data extraction |
NuNER-multilingual | HuggingFace | Entity-Recognition | * | 711Mb | Maybe, entity-extraction in newspapers |
NuNER-Zero | HuggingFace | Zero-Shot Named-Entity-Recognition | * | 1.8Gb | Fast new-shot entity extraction |
NuTopic | HuggingFace | Topic-Recognition | * | 438Mb | Fast, topic-extraction |
TimesFM | HuggingFace | Time-Series Foundation Models | 200M | 814Mb | Maybe, zero-shot anomaly detection |
LittleTinies | HuggingFace | Text-to-Cartoonish-Image-Generation | * | 228Mb | Fast, cutesy storyboarding |
Storyboard-Sketch | HuggingFace | Text-to-Storyboard-Sketch | * | 228Mb | Fast, sketch plot teller |
ShieldGemma | HuggingFace | Content Moderation | 2.61B params | 5.2Gb | High-performance safety content moderation in text generation applications |