Sujit Vasanth PRO
sujitvasanth
·
AI & ML interests
inference optimisation, gait reinforcement learning for mobile robots (quadrupeds, bipeds), song (music and vocal) generation, local computer use models and agents
Organizations
None yet
models performance sats is for the 31b
4
#1 opened 3 months ago
by sujitvasanth
respectfully: expanded CI suite breaks all repo github builds
#3 opened 2 months ago
by sujitvasanth
imatrix gguf's
2
#1 opened 3 months ago
by sujitvasanth
google/gemma-4-26B-A4B-it-assistant
👍 2
3
#1 opened 3 months ago
by sujitvasanth
Hcompany/Holo2-4B
1
#10 opened 8 months ago
by sujitvasanth
Qwen3VL-8B with WORKING coordinates please!
1
#9 opened 8 months ago
by sujitvasanth
Great model!
#1 opened 9 months ago
by sujitvasanth
Nice voice!
#1 opened 9 months ago
by sujitvasanth
Great voice!
#1 opened 9 months ago
by sujitvasanth
Wow so fast!
3
#1 opened 10 months ago
by sujitvasanth
I have produced quantized exllamav2 version 6Gb and much faster inference
5
#3 opened 10 months ago
by sujitvasanth
Model uses server not your huggingface model
#1 opened 11 months ago
by sujitvasanth
Grounding task always (incorrectly) produce click action
#2 opened 11 months ago
by sujitvasanth
Practicalities of fast inference (NOT OSWorld!)
1
#2 opened 11 months ago
by sujitvasanth
add pipeline
9
#6 opened over 2 years ago
by 
not-lain
Example of tinyllava model implementation Tk transformers
2
#2 opened over 2 years ago
by sujitvasanth
The model supports multi-image and multi-prompt generation.?
2
#1 opened over 2 years ago
by sujitvasanth
Quants
7
#1 opened over 2 years ago
by 
CyberTimon