INDEX

Explanations

semicolons used in long lists or explanations

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, particularly semicolons

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ema

-0.76

edo

-0.73

ħĭ

-0.71

stration

-0.68

frac

-0.66

inki

-0.66

IDENT

-0.66

burgh

-0.65

omorphic

-0.65

se

-0.64

POSITIVE LOGITS

 moreover

1.46

 furthermore

1.39

 however

1.25

 meanwhile

1.24

 consequently

1.22

 hence

1.21

 nevertheless

1.18

 alternatively

1.17

 indeed

1.13

 namely

1.12

Activations Density 0.075%

semicolons used in long lists or explanations

punctuation marks, particularly semicolons

No Comments

No Known Activations