INDEX

Explanations

descriptions or titles related to rankings or acknowledgments

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate rankings or distinctions of people or things

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

icates

-0.76

parts

-0.65

Results

-0.65

anol

-0.62

mare

-0.61

cffff

-0.60

 pauses

-0.58

 attaching

-0.58

icated

-0.57

osponsors

-0.57

POSITIVE LOGITS

 finest

1.41

 greatest

1.38

 smartest

1.36

 richest

1.34

 coolest

1.28

 wealthiest

1.27

 toughest

1.26

 fastest

1.23

 hottest

1.21

 deadliest

1.20

Activations Density 0.149%

descriptions or titles related to rankings or acknowledgments

phrases that indicate rankings or distinctions of people or things

No Comments

No Known Activations