INDEX

Explanations

phrases indicating potential actions, recommendations, or considerations

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "we" and its variants indicating collective thoughts or actions

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

amaz

-0.62

Rank

-0.60

dylib

-0.58

bats

-0.58

rams

-0.56

shows

-0.54

 Cheong

-0.54

 satell

-0.54

WTC

-0.54

Ups

-0.53

POSITIVE LOGITS

 sorely

1.13

 dearly

0.90

 aspire

0.87

 gladly

0.86

 dreamed

0.85

 envy

0.81

 wont

0.79

 desperately

0.79

 vehemently

0.78

 hotly

0.78

Activations Density 0.155%

phrases indicating potential actions, recommendations, or considerations

instances of the phrase "we" and its variants indicating collective thoughts or actions

No Comments

No Known Activations