INDEX

Explanations

phrases related to controversial topics or debates

oai_token-act-pair · gpt-3.5-turbo

occurrences of notation or symbols related to lists or references within a text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 cradle

-0.74

awaru

-0.70

aeus

-0.68

dy

-0.68

ts

-0.67

anx

-0.67

icals

-0.65

tin

-0.65

roe

-0.64

otin

-0.64

POSITIVE LOGITS

CLASSIFIED

1.09

...]

1.04

âĢ¦]

0.99

 [...]

0.92

 Appears

0.90

externalActionCode

0.90

},{"

0.89

ÂłÂłÂłÂłÂłÂłÂłÂł

0.89

âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦

0.88

 [âĢ¦]

0.86

Activations Density 0.004%

phrases related to controversial topics or debates

occurrences of notation or symbols related to lists or references within a text

No Comments

No Known Activations