INDEX

Explanations

specific information related to correspondence in the form of letters

oai_token-act-pair · gpt-3.5-turbo

instances of letters and correspondence in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

 eyed

-0.71

 ponds

-0.69

 spectator

-0.69

é¾

-0.68

Course

-0.66

 parks

-0.66

hog

-0.65

 lane

-0.65

 stadiums

-0.64

OVA

-0.63

POSITIVE LOGITS

Dear

1.01

 dated

1.00

Letter

0.95

 outlining

0.92

 Letters

0.91

 correspondence

0.89

 letter

0.89

 letters

0.88

PsyNetMessage

0.88

 informing

0.87

Activations Density 0.363%

specific information related to correspondence in the form of letters

instances of letters and correspondence in the text

No Comments

No Known Activations