INDEX

Explanations

mentions of speakers making statements or expressing opinions

oai_token-act-pair · gpt-3.5-turbo

statements attributed to different speakers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ugal

-0.87

estern

-0.82

folios

-0.72

 earthqu

-0.68

nerg

-0.68

ĸļå£«

-0.68

 pione

-0.68

ÃĥÃĤÃĥÃĤ

-0.66

ACTED

-0.65

ortunately

-0.65

POSITIVE LOGITS

 goodbye

0.89

omething

0.78

upon

0.75

 STATS

0.70

ynthesis

0.68

 says

0.65

tics

0.65

 Merry

0.64

 Birch

0.63

andra

0.63

Activations Density 0.078%

mentions of speakers making statements or expressing opinions

statements attributed to different speakers

No Comments

No Known Activations