INDEX

Explanations

political terms related to governance or policies

oai_token-act-pair · gpt-3.5-turbo

references to political terms and concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

olescent

-0.64

 SCHOOL

-0.64

 standpoint

-0.63

illet

-0.63

lad

-0.63

umatic

-0.60

ometric

-0.59

 AUTHOR

-0.59

 LIMITED

-0.58

onica

-0.58

POSITIVE LOGITS

 together

1.06

together

0.90

 Together

0.87

NSA

0.72

————————————————

0.72

Sov

0.71

oran

0.70

Together

0.70

terms

0.69

ãģ®é

0.67

Activations Density 0.000%

political terms related to governance or policies

references to political terms and concepts

No Comments

No Known Activations