INDEX

Explanations

specific numerical quantities related to people or objects mentioned in the text

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

urden

-0.84

obin

-0.84

urgy

-0.76

vantage

-0.75

Platform

-0.74

////

-0.73

vention

-0.71

answer

-0.70

ģ«

-0.70

à¼

-0.70

POSITIVE LOGITS

 unnamed

1.11

 prominent

1.09

 dozen

1.02

 unidentified

0.98

 teenage

0.97

 individuals

0.94

 brothers

0.94

men

0.93

 teenagers

0.93

 former

0.92

Activations Density 0.180%

specific numerical quantities related to people or objects mentioned in the text

No Comments

No Known Activations