INDEX

Explanations

phrases indicating a high cost or negative consequence

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the cost or consequences of actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

plane

-0.64

valid

-0.64

activate

-0.63

planes

-0.61

FTWARE

-0.60

available

-0.59

Lens

-0.58

 WRITE

-0.58

podcast

-0.57

Phill

-0.57

POSITIVE LOGITS

 least

1.09

 halftime

0.94

onement

0.83

 odds

0.80

roph

0.73

 midnight

0.72

itud

0.71

 speeds

0.70

 midday

0.69

 logger

0.69

Activations Density 0.094%

phrases indicating a high cost or negative consequence

phrases indicating the cost or consequences of actions

No Comments

No Known Activations