This is a collection of Benchmarking results between Gradio and FastMCP
Kshitij Thakkar PRO
kshitijthakkar
AI & ML interests
AI observability + MoE efficiency engineer. Building tools that make GenAI traceable, measurable, and production-ready.
Recent Activity
liked a Space 2 days ago
smolagents/ml-intern updated a Space 4 days ago
kshitijthakkar/guardiantails updated a Space 6 days ago
kshitijthakkar/guardiantailsOrganizations
Systematic search for 1B-2B MoE models. Best: bs=1, ctx=2048 achieves 0.32 loss. Top-8 routing beats top-2.
My submission to Agents-MCP-Hackathon
Collection of Loggenix Models, Eval Dataset, Demo Playground. Soon will add the training dataset.
-
Loggenix Moe 0.3B A0.1B Demo
🏢2Demo Space for my model loggenix-moe-0.3B-A0.1B
-
kshitijthakkar/loggenix-synthetic-ai-tasks-eval-with-outputs
Viewer • Updated • 28 • 8 -
kshitijthakkar/loggenix-synthetic-ai-tasks-eval_v6-with-outputs
Viewer • Updated • 170 • 6 -
kshitijthakkar/loggenix-synthetic-ai-tasks-eval_v5-with-outputs-v7-sft-v1
Viewer • Updated • 170 • 14
Qwen3.5 MoE models from dual-source weight transfer (dense backbone + 35B-A3B experts). Hybrid DeltaNet + GQA attention.
-
kshitijthakkar/qwen3.5-moe-0.87B-d0.8B
Image-Text-to-Text • 1B • Updated • 13 -
kshitijthakkar/qwen3.5-moe-2.3B-d2B
Image-Text-to-Text • 3B • Updated • 10 -
kshitijthakkar/qwen3.5-moe-4.7B-d4B
Image-Text-to-Text • 5B • Updated • 36 -
kshitijthakkar/qwen3.5-tiny-test
Image-Text-to-Text • 0.1B • Updated • 14
32 MoE models from 41 experiments exploring expert count, routing, and learning rates for mobile deployment.
Collection of my submissions to MCP-1st-Birthday:- TraceMind Agent and MCP Server and smoltrace datasets generated for running evals using smoltrace.
This is a collection of Benchmarking results between Gradio and FastMCP
Qwen3.5 MoE models from dual-source weight transfer (dense backbone + 35B-A3B experts). Hybrid DeltaNet + GQA attention.
-
kshitijthakkar/qwen3.5-moe-0.87B-d0.8B
Image-Text-to-Text • 1B • Updated • 13 -
kshitijthakkar/qwen3.5-moe-2.3B-d2B
Image-Text-to-Text • 3B • Updated • 10 -
kshitijthakkar/qwen3.5-moe-4.7B-d4B
Image-Text-to-Text • 5B • Updated • 36 -
kshitijthakkar/qwen3.5-tiny-test
Image-Text-to-Text • 0.1B • Updated • 14
Systematic search for 1B-2B MoE models. Best: bs=1, ctx=2048 achieves 0.32 loss. Top-8 routing beats top-2.
32 MoE models from 41 experiments exploring expert count, routing, and learning rates for mobile deployment.
My submission to Agents-MCP-Hackathon
Collection of my submissions to MCP-1st-Birthday:- TraceMind Agent and MCP Server and smoltrace datasets generated for running evals using smoltrace.
Collection of Loggenix Models, Eval Dataset, Demo Playground. Soon will add the training dataset.
-
Loggenix Moe 0.3B A0.1B Demo
🏢2Demo Space for my model loggenix-moe-0.3B-A0.1B
-
kshitijthakkar/loggenix-synthetic-ai-tasks-eval-with-outputs
Viewer • Updated • 28 • 8 -
kshitijthakkar/loggenix-synthetic-ai-tasks-eval_v6-with-outputs
Viewer • Updated • 170 • 6 -
kshitijthakkar/loggenix-synthetic-ai-tasks-eval_v5-with-outputs-v7-sft-v1
Viewer • Updated • 170 • 14