INDEX

Explanations

news-related terms in sentences

oai_token-act-pair · gpt-3.5-turbo

instances of the word "reported," indicating references to news or announcements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

isable

-0.79

creen

-0.71

venge

-0.69

ophone

-0.67

youtube

-0.64

gencies

-0.64

 gauge

-0.64

ophers

-0.63

uv

-0.63

ococ

-0.63

POSITIVE LOGITS

 quoting

0.84

 extensively

0.74

 yesterday

0.74

 ARTICLE

0.70

 excerpts

0.69

biz

0.68

 citing

0.67

 reporter

0.67

 headlined

0.67

 reporters

0.67

Activations Density 0.048%

news-related terms in sentences

instances of the word "reported," indicating references to news or announcements

No Comments

No Known Activations