INDEX

Explanations

words related to political discussions and policies

oai_token-act-pair · gpt-3.5-turbo

references to political conflicts and crises

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

\">

-0.79

PokÃ©

-0.78

\",

-0.76

"))

-0.74

ãĢį

-0.72

Currently

-0.69

?".

-0.69

ADVERTISEMENT

-0.69

VERTISEMENT

-0.68

 INFO

-0.68

POSITIVE LOGITS

 repud

1.13

cyn

0.99

 doubtless

0.97

 scarcely

0.94

 cynicism

0.94

 indign

0.90

 capit

0.90

 folly

0.88

 shrew

0.88

iberal

0.88

Activations Density 1.831%

words related to political discussions and policies

references to political conflicts and crises

No Comments

No Known Activations