INDEX

Explanations

political news and events, particularly accusations and actions involving various countries and individuals

oai_token-act-pair · gpt-3.5-turbo

references to political blame and accountability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Operator

-0.83

 UNIVERS

-0.78

(?,

-0.76

ortun

-0.66

 Instruction

-0.65

NON

-0.65

 Solution

-0.65

 Folder

-0.63

 glim

-0.63

TTL

-0.63

POSITIVE LOGITS

 denounced

0.94

 denouncing

0.93

 accuse

0.90

 accusing

0.87

 staunch

0.87

 discredited

0.86

 criticized

0.86

 scorn

0.86

 vehemently

0.86

 criticizing

0.85

Activations Density 0.832%

political news and events, particularly accusations and actions involving various countries and individuals

references to political blame and accountability

No Comments

No Known Activations