INDEX

Explanations

phrases or sentences marked by the word "Interestingly"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Interestingly" or its variations, suggesting a focus on statements that introduce surprising or notable information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 newsletters

-0.87

needs

-0.63

 directions

-0.63

"},"

-0.62

 GOLD

-0.62

estern

-0.61

 postage

-0.59

bern

-0.57

 processing

-0.57

iche

-0.57

POSITIVE LOGITS

Interestingly

0.92

 interestingly

0.85

 Interestingly

0.85

theless

0.82

REDACTED

0.75

mobi

0.73

bley

0.73

 Ironically

0.72

tenance

0.72

SPONSORED

0.72

Activations Density 0.013%

phrases or sentences marked by the word "Interestingly"

instances of the word "Interestingly" or its variations, suggesting a focus on statements that introduce surprising or notable information

No Comments

No Known Activations