INDEX

Explanations

mentions of specific locations or events, possibly related to news articles

oai_token-act-pair · gpt-3.5-turbo

sections of text that contain lists or sequential information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 endeavour

-0.66

 scrap

-0.65

 endeav

-0.65

 realised

-0.63

 undermin

-0.61

utical

-0.59

 freel

-0.59

 deceived

-0.59

 answ

-0.59

utics

-0.57

POSITIVE LOGITS

ÂłÂłÂłÂł

0.88

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.86

ÂłÂłÂłÂłÂłÂłÂłÂł

0.85

Advertisement

0.78

Loading

0.75

Article

0.74

THIS

0.74

Govern

0.73

ÂłÂłÂł

0.73

Notice

0.71

Activations Density 1.027%

mentions of specific locations or events, possibly related to news articles

sections of text that contain lists or sequential information

No Comments

No Known Activations