INDEX

Explanations

phrases related to discussing, proposing, or evaluating ideas or plans

oai_token-act-pair · gpt-3.5-turbo

references to significant ideas or concepts

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

record

-0.80

 audits

-0.78

 Coverage

-0.76

essional

-0.75

 penalties

-0.73

adelphia

-0.72

 Penalty

-0.71

 Records

-0.70

reporting

-0.70

LIA

-0.69

POSITIVE LOGITS

 floated

1.18

 hatched

1.17

 appealing

0.98

 revived

0.97

 gaining

0.95

 embraced

0.95

 entertained

0.95

 prevalent

0.94

 debunked

0.94

 concept

0.94

Activations Density 0.245%

phrases related to discussing, proposing, or evaluating ideas or plans

references to significant ideas or concepts

No Comments

No Known Activations