INDEX

Explanations

phrases related to news sources and publications

oai_token-act-pair · gpt-3.5-turbo

references to specific organizations, publications, and websites

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

agos

-0.75

etts

-0.73

ivas

-0.69

plot

-0.66

hend

-0.66

say

-0.66

aceae

-0.64

Äį

-0.62

sei

-0.61

ages

-0.61

POSITIVE LOGITS

ermanent

0.59

 Subscribe

0.59

 subscribers

0.58

 subscription

0.58

 contributors

0.58

 Weekly

0.57

Magazine

0.57

Month

0.56

culosis

0.55

ribly

0.55

Activations Density 0.110%

phrases related to news sources and publications

references to specific organizations, publications, and websites

No Comments

No Known Activations