Human-Intelligence/storage
6.74 MB
AI & ML interests
Humans answer your questions.
Recent Activity
wopΒ
updated a
bucket 1 day ago
Post
149
π __Monostep v1__ is up β[Monostep-v1 Demo}( wop/Cosmos-T2-Chat)
A tiny (~16.6M) experimental model that predicts 4 tokens per forward pass instead of one. A Transformer trunk pools the prompt into a single vector, then 4 sequential "slot" heads emit a block of tokens left-to-right β a lightweight take on multi-token prediction.
Trained on GSM8K (GPT-2 tokenizer, 10 epochs). It's small and rough β answers are often wrong β but it's a fun little testbed for block decoding. Weights, config, training curves, and a self-contained inference snippet are all in the repo.
Also wired into the Cosmos T2-Accelerate chat demo, where it streams those 4-token blocks live. π§ͺ
#multitokenprediction #gsm8k #smallmodels
A tiny (~16.6M) experimental model that predicts 4 tokens per forward pass instead of one. A Transformer trunk pools the prompt into a single vector, then 4 sequential "slot" heads emit a block of tokens left-to-right β a lightweight take on multi-token prediction.
Trained on GSM8K (GPT-2 tokenizer, 10 epochs). It's small and rough β answers are often wrong β but it's a fun little testbed for block decoding. Weights, config, training curves, and a self-contained inference snippet are all in the repo.
Also wired into the Cosmos T2-Accelerate chat demo, where it streams those 4-token blocks live. π§ͺ
#multitokenprediction #gsm8k #smallmodels
wopΒ
updated 2
Spaces 27 days ago
wopΒ
Update templates/index.html
#2 opened 28 days ago
by 
wop
wopΒ
updated a
Space about 1 month ago
wopΒ
published a
Space about 1 month ago
wopΒ
Update templates/index.html
2
#1 opened about 1 month ago
by 
CompactAI
wopΒ
published a
bucket about 1 month ago
wopΒ
published a
Space about 1 month ago
wopΒ
updated a
bucket about 1 month ago
wopΒ
published a
Space about 1 month ago
wopΒ
updated a
Space about 1 month ago
wopΒ
published a
bucket about 1 month ago
wopΒ
published a
Space about 1 month ago