INDEX

Explanations

parenthetical information

oai_token-act-pair · gpt-3.5-turbo

opening parentheses, indicating various subsections or references within the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 inund

-0.79

 everyday

-0.77

 obscurity

-0.73

 drifting

-0.72

ris

-0.71

 absor

-0.70

 patri

-0.70

 tackling

-0.70

 veget

-0.69

 lull

-0.69

POSITIVE LOGITS

emphasis

1.17

âĢ¦)

1.13

...)

1.11

laughs

1.07

Note

1.06

See

1.02

aka

0.99

Photo

0.99

++)

0.98

CBC

0.97

Activations Density 0.100%

parenthetical information

opening parentheses, indicating various subsections or references within the text

No Comments

No Known Activations