Article
Testing selected-token J-lens directions and a fixed-symbol autoencoder on Qwen3-8B
kotlarmilos
• Milos Kotlar PRO
kotlarmilos
webbrain-one-404441's profile picture
melihakbulut221's profile picture
webbrain-one-397620's profile picture·
AI & ML interests
None yet
Recent Activity
published an article about 1 month ago
Testing selected-token J-lens directions and a fixed-symbol autoencoder on Qwen3-8B published an article about 1 month ago
Predicting GPU communication time from measured collectives published an article about 1 month ago
Testing early stopping in a diffusion language modelOrganizations
None yet
published an article about 1 month ago
published an article about 1 month ago
Article
Predicting GPU communication time from measured collectives
kotlarmilos
• published an article about 1 month ago
Article
Testing early stopping in a diffusion language model
kotlarmilos
• updated 3
models about 1 month ago
published 3
models about 1 month ago
published an article about 1 month ago
Article
Reducing dropped tokens in a small MoE model
kotlarmilos
• updated a
model about 1 month ago
published a
model about 1 month ago
published an article about 1 month ago
Article
Allocating latent KV-cache memory across transformer layers
kotlarmilos
• updated a
model about 1 month ago
published a
model about 1 month ago
published an article about 1 month ago
Article
Shrinking my transformer's KV cache with a KL calibration budget
kotlarmilos
• updated a
model about 1 month ago
updated a
model 4 months ago
published a
model 4 months ago
published an article 4 months ago
Article
Teaching Gemma 4 to predict glucose from a meal
kotlarmilos
•