INDEX

Explanations

instances of the word "elated" or variations of it

oai_token-act-pair · gpt-3.5-turbo

instances of the word "eluded."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Accountability

-0.63

 shatter

-0.61

stress

-0.61

 Continued

-0.59

 Rolls

-0.58

Lov

-0.58

 Affect

-0.56

 rocking

-0.56

ãĥ´ãĤ¡

-0.55

NEY

-0.55

POSITIVE LOGITS

izabeth

1.30

ixir

1.27

uded

1.21

itism

1.15

uding

1.07

iquid

1.06

usive

1.05

abor

1.02

ijah

1.01

apses

0.93

Activations Density 0.019%

instances of the word "elated" or variations of it

instances of the word "eluded."

No Comments

No Known Activations