INDEX

Explanations

numerical information related to data analysis and tracking

oai_token-act-pair · gpt-3.5-turbo

metrics related to measurements and data tracking

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

¿½

-0.65

©¶æ

-0.63

exclusive

-0.62

pret

-0.61

ahon

-0.59

arlane

-0.59

 Airl

-0.58

elected

-0.58

evil

-0.58

annon

-0.57

POSITIVE LOGITS

 accordingly

1.15

 beforehand

1.05

etc

1.01

 accurately

1.00

 BEFORE

0.97

 pertinent

0.96

 aloud

0.92

 before

0.92

 whenever

0.89

 whereabouts

0.87

Activations Density 0.505%

numerical information related to data analysis and tracking

metrics related to measurements and data tracking

No Comments

No Known Activations