INDEX

Explanations

phrases related to surprising or strange facts or events

oai_token-act-pair · gpt-3.5-turbo

phrases that express surprising or unexpected outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

robe

-0.75

irie

-0.74

 guiActiveUnfocused

-0.72

enhagen

-0.70

gang

-0.70

ļéĨĴ

-0.66

iami

-0.65

enaries

-0.64

eria

-0.62

ioxide

-0.62

POSITIVE LOGITS

 especially

1.04

huh

1.02

 though

1.01

 albeit

0.98

 considering

0.98

but

0.98

 however

0.95

 although

0.94

but

0.90

 because

0.88

Activations Density 0.288%

phrases related to surprising or strange facts or events

phrases that express surprising or unexpected outcomes

No Comments

No Known Activations