INDEX

Explanations

descriptions or mentions of jobs or employment status, specifically focusing on full-time positions

oai_token-act-pair · gpt-3.5-turbo

references to full-time employment

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

ĸļ

-0.94

hra

-0.84

rosis

-0.83

ovember

-0.75

ambo

-0.73

ramid

-0.71

reth

-0.71

acid

-0.69

romeda

-0.69

Ì¶

-0.69

POSITIVE LOGITS

 employment

0.98

 equivalents

0.96

 basis

0.75

 missionary

0.73

 worker

0.70

 unemployed

0.70

 income

0.69

 missionaries

0.68

 occupation

0.68

 equivalent

0.67

Activations Density 0.044%

descriptions or mentions of jobs or employment status, specifically focusing on full-time positions

references to full-time employment

No Comments

No Known Activations