INDEX

Explanations

contexts where something is described as being superior, elevated, or standing out

oai_token-act-pair · gpt-3.5-turbo

instances of the word "above" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

dll

-0.70

 Instruments

-0.69

Introduced

-0.68

 Centers

-0.66

pring

-0.65

ately

-0.65

locks

-0.64

Roh

-0.63

LOCK

-0.63

 Sacrament

-0.63

POSITIVE LOGITS

perty

0.96

ground

0.89

mentioned

0.79

uckland

0.76

IFA

0.73

ceed

0.70

raf

0.69

uve

0.69

 Above

0.66

lihood

0.65

Activations Density 0.018%

contexts where something is described as being superior, elevated, or standing out

instances of the word "above" in various contexts

No Comments

No Known Activations