INDEX

Explanations

words related to secret information or investigations

oai_token-act-pair · gpt-3.5-turbo

references to investigative reports or documents containing allegations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

speed

-0.81

Pyr

-0.76

rh

-0.76

rl

-0.75

Saw

-0.69

Wilson

-0.68

 Mariners

-0.67

alone

-0.65

gro

-0.65

 Instrument

-0.64

POSITIVE LOGITS

 dossier

1.46

 trove

0.96

 compiled

0.87

ossier

0.83

 bombshell

0.83

 memos

0.79

 fodder

0.75

 allegations

0.74

 Ukrain

0.73

 alleging

0.73

Activations Density 0.027%

words related to secret information or investigations

references to investigative reports or documents containing allegations

No Comments

No Known Activations