INDEX

Explanations

quantitative adverbs (e.g., quite, fairly, significantly) followed by adjectives expressing degrees of probability

oai_token-act-pair · gpt-3.5-turbo

adverbs that indicate degree or intensity

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Flavoring

-1.12

rosis

-0.90

 Awakens

-0.83

ngth

-0.77

 openings

-0.75

ortmund

-0.73

cue

-0.72

rals

-0.71

enaries

-0.71

llor

-0.71

POSITIVE LOGITS

 dependent

1.08

 reliant

1.06

 evident

1.05

 accurate

1.02

 accessible

1.01

 prevalent

1.00

 correlated

0.99

 irrelevant

0.99

 profitable

0.98

 useful

0.97

Activations Density 0.337%

quantitative adverbs (e.g., quite, fairly, significantly) followed by adjectives expressing degrees of probability

adverbs that indicate degree or intensity

No Comments

No Known Activations