INDEX

Explanations

Keywords related to political figures and events

oai_token-act-pair · gpt-3.5-turbo

references to political figures and their affiliations

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

 civilisation

-0.81

ĸļ

-0.79

 mould

-0.78

 realised

-0.77

 metre

-0.76

 realise

-0.75

 organise

-0.74

 analyse

-0.73

 recognised

-0.73

 Whilst

-0.73

POSITIVE LOGITS

Republicans

1.31

GOP

1.29

Democrats

1.28

 Politico

1.24

Hillary

1.24

 Republican

1.22

Republican

1.22

Obama

1.22

Trump

1.20

 Democrats

1.20

Activations Density 2.233%

Keywords related to political figures and events

references to political figures and their affiliations

No Comments

No Known Activations