INDEX

Explanations

phrases starting with "When" to indicate a sequence of events

oai_token-act-pair · gpt-3.5-turbo

instances of the word "When" to indicate events in a narrative context

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

agin

-0.76

zag

-0.65

abal

-0.63

\\\\\\\\

-0.62

roman

-0.61

udder

-0.61

JV

-0.61

Í

-0.60

aido

-0.60

åĤ

-0.60

POSITIVE LOGITS

 asked

1.22

 confronted

1.12

 questioned

1.10

 pressed

1.08

soever

1.07

 contacted

1.01

 quizz

0.91

 approached

0.85

 faced

0.85

 interviewer

0.82

Activations Density 0.079%

phrases starting with "When" to indicate a sequence of events

instances of the word "When" to indicate events in a narrative context

No Comments

No Known Activations