INDEX

Explanations

terms related to credibility or incredulity

oai_token-act-pair · gpt-3.5-turbo

terms related to credibility and belief

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

meal

-0.76

Ò

-0.76

hap

-0.74

Won

-0.67

Bed

-0.66

 Boat

-0.64

 Nurs

-0.63

 Eden

-0.62

 Abandon

-0.61

 Result

-0.61

POSITIVE LOGITS

enza

1.32

ulously

1.32

ibly

1.31

ulous

1.25

ulent

1.04

entials

0.99

icator

0.99

iosity

0.98

ulum

0.97

iting

0.96

Activations Density 0.007%

terms related to credibility or incredulity

terms related to credibility and belief

No Comments

No Known Activations