INDEX

Explanations

phrases indicating information or actions being attributed to a particular source or authority

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

âĻ¦

-0.62

ql

-0.62

 realise

-0.61

eg

-0.59

 heights

-0.59

blast

-0.59

 roller

-0.57

apons

-0.56

swe

-0.55

apes

-0.55

POSITIVE LOGITS

ly

1.10

edly

0.91

resa

0.82

translation

0.73

tained

0.73

itionally

0.72

LY

0.71

edIn

0.70

lies

0.70

ificantly

0.70

Activations Density 0.036%

phrases indicating information or actions being attributed to a particular source or authority

No Comments

No Known Activations