INDEX

Explanations

phrases related to reported or rumored actions or situations

oai_token-act-pair · gpt-3.5-turbo

the term "reportedly" indicating news or statements that are based on hearsay or unverified information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

cest

-0.74

uit

-0.71

ados

-0.67

bern

-0.66

lap

-0.65

ces

-0.65

eral

-0.64

onest

-0.63

ravel

-0.63

gers

-0.63

POSITIVE LOGITS

 located

0.77

Ĥİ

0.74

 priced

0.69

 reportedly

0.66

¶

0.66

 obtained

0.63

 held

0.61

 purchased

0.61

 proport

0.61

 owned

0.61

Activations Density 0.016%

phrases related to reported or rumored actions or situations

the term "reportedly" indicating news or statements that are based on hearsay or unverified information

No Comments

No Known Activations