INDEX

Explanations

references to political figures and events

oai_token-act-pair · gpt-3.5-turbo

political discourse and references to policies

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

 metre

-0.75

 realise

-0.72

 realised

-0.71

 endeavour

-0.70

 analyse

-0.70

gements

-0.70

 explorers

-0.69

 organise

-0.68

 labou

-0.67

 mould

-0.67

POSITIVE LOGITS

 Politico

1.13

 Republican

1.12

GOP

1.08

 bipartisan

1.07

 Hillary

1.06

partisan

1.02

 congressional

1.01

Republican

1.00

Republicans

0.99

 Biden

0.99

Activations Density 3.162%

references to political figures and events

political discourse and references to policies

No Comments

No Known Activations