INDEX

Explanations

words related to leadership positions or titles

oai_token-act-pair · gpt-3.5-turbo

references to military leadership positions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

eat

-0.75

aver

-0.71

-0.65

alle

-0.64

employment

-0.63

ropy

-0.63

encers

-0.63

apy

-0.62

ted

-0.61

Democrats

-0.60

POSITIVE LOGITS

chief

0.82

 commands

0.77

 commanding

0.77

stration

0.76

 commander

0.76

ials

0.74

ially

0.73

ufact

0.72

itatively

0.72

 commanders

0.70

Activations Density 0.012%

words related to leadership positions or titles

references to military leadership positions

No Comments

No Known Activations