INDEX

Explanations

information related to news stories and events

oai_token-act-pair · gpt-3.5-turbo

references to political statements and actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

uzz

-0.68

 Darius

-0.68

aco

-0.68

 Â¯

-0.66

 Secondly

-0.63

Eag

-0.62

reddit

-0.62

DC

-0.61

Princ

-0.61

 Spicer

-0.61

POSITIVE LOGITS

 careers

0.91

 themselves

0.82

osponsors

0.80

 respective

0.79

 spouses

0.79

 either

0.76

 majors

0.74

 varying

0.73

 specialize

0.73

 divest

0.71

Activations Density 1.248%

information related to news stories and events

references to political statements and actions

No Comments

No Known Activations