INDEX

Explanations

phrases related to advising or warning others

oai_token-act-pair · gpt-3.5-turbo

references to relationships and communal support

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ital

-0.62

igun

-0.61

english

-0.60

geoning

-0.59

 guiActiveUnfocused

-0.59

 Investig

-0.58

ensus

-0.57

adal

-0.56

osta

-0.56

ettel

-0.55

POSITIVE LOGITS

 others

1.12

 yours

0.92

 theirs

0.90

 coworkers

0.84

 ours

0.82

 Associates

0.80

 Others

0.78

 hers

0.77

 mine

0.77

romeda

0.73

Activations Density 0.089%

phrases related to advising or warning others

references to relationships and communal support

No Comments

No Known Activations