INDEX

Explanations

information about individuals or groups leading an effort or project

oai_token-act-pair · gpt-3.5-turbo

instances of the word "led" in contexts indicating leadership or direction

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

 Scotia

-0.76

 sunset

-0.76

 Ratio

-0.61

 Addiction

-0.60

 Smoking

-0.59

 Buildings

-0.59

 breeze

-0.58

Done

-0.57

iddler

-0.57

cube

-0.57

POSITIVE LOGITS

gers

1.19

gling

1.03

ges

0.95

better

0.94

ging

0.93

ger

0.88

ged

0.84

leaders

0.80

led

0.79

ership

0.79

Activations Density 0.031%

information about individuals or groups leading an effort or project

instances of the word "led" in contexts indicating leadership or direction

No Comments

No Known Activations