INDEX

Explanations

phrases mentioning a wide range of topics or subjects

oai_token-act-pair · gpt-3.5-turbo

references to various subjects or concepts described as "things."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

esi

-0.82

idav

-0.77

imens

-0.72

ardo

-0.70

xon

-0.70

bern

-0.70

aram

-0.70

oir

-0.66

ramids

-0.66

gee

-0.65

POSITIVE LOGITS

 imaginable

0.95

 pertaining

0.88

 transpired

0.82

 else

0.81

 happening

0.80

 relating

0.79

Else

0.75

 happened

0.72

 happens

0.69

afety

0.69

Activations Density 0.048%

phrases mentioning a wide range of topics or subjects

references to various subjects or concepts described as "things."

No Comments

No Known Activations