INDEX

Explanations

references to news reports or coverage

oai_token-act-pair · gpt-3.5-turbo

instances of the word "report" and its variations in the text

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

ĸļ

-0.83

ophers

-0.73

antics

-0.73

uit

-0.70

conservancy

-0.69

etsk

-0.68

rongh

-0.67

uits

-0.66

 palate

-0.65

ophone

-0.64

POSITIVE LOGITS

è¦ļéĨĴ

0.89

 quoting

0.84

 reports

0.83

age

0.83

 sources

0.80

 quotes

0.80

WASHINGTON

0.77

 rumors

0.76

 leaked

0.76

 quoted

0.75

Activations Density 0.055%

references to news reports or coverage

instances of the word "report" and its variations in the text

No Comments

No Known Activations