INDEX

Explanations

mentions of military ranks, specifically the abbreviation "Lt" followed by a number

oai_token-act-pair · gpt-3.5-turbo

titles or ranks related to military or governmental positions

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

BOOK

-0.87

FORE

-0.80

Offline

-0.79

stakes

-0.77

âĶģ

-0.75

anwhile

-0.67

raised

-0.67

theless

-0.67

ãĥ¼ãĥ³

-0.67

*/(

-0.66

POSITIVE LOGITS

 colonel

1.09

utenant

1.03

 Colonel

0.93

Lt

0.84

Lt

0.84

Col

0.82

seless

0.80

 Commander

0.79

geant

0.74

 lieutenant

0.73

Activations Density 0.012%

mentions of military ranks, specifically the abbreviation "Lt" followed by a number

titles or ranks related to military or governmental positions

No Comments

No Known Activations