INDEX

Explanations

phrases related to positions or titles of authority, particularly those referring to leaders or heads

oai_token-act-pair · gpt-3.5-turbo

references to leadership roles and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

PsyNetMessage

-0.91

ickr

-0.78

 mell

-0.78

 resil

-0.77

Avg

-0.67

 volunt

-0.65

 bottleneck

-0.64

Constructed

-0.64

 chloride

-0.63

 succeeding

-0.59

POSITIVE LOGITS

phones

1.18

quartered

1.09

ache

1.07

butt

1.03

quarters

1.02

quarter

1.01

lining

1.00

waters

1.00

gear

0.98

liners

0.97

Activations Density 0.010%

phrases related to positions or titles of authority, particularly those referring to leaders or heads

references to leadership roles and titles

No Comments

No Known Activations