INDEX

Explanations

special characters that look like a combination of letters or symbols

oai_token-act-pair · gpt-3.5-turbo

instances of high numerical values or significant emphasis

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

heit

-0.83

boro

-0.79

abouts

-0.74

ounce

-0.65

ennes

-0.61

itude

-0.61

ocial

-0.61

utical

-0.60

itory

-0.58

anqu

-0.57

POSITIVE LOGITS

reditary

0.78

ÂłÂłÂł

0.78

 Debor

0.70

ÂłÂłÂłÂłÂłÂłÂłÂł

0.65

However

0.65

Lear

0.64

ÂłÂłÂłÂł

0.63

ECK

0.62

Scroll

0.61

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.61

Activations Density 0.707%

special characters that look like a combination of letters or symbols

instances of high numerical values or significant emphasis

No Comments

No Known Activations