INDEX

Explanations

proper nouns related to roles, names, and titles

oai_token-act-pair · gpt-3.5-turbo

references to leadership or authoritative figures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

PsyNetMessage

-0.91

ickr

-0.79

 mell

-0.77

 resil

-0.71

Avg

-0.68

vo

-0.65

 suppl

-0.65

 chloride

-0.63

ktop

-0.62

 bottleneck

-0.61

POSITIVE LOGITS

quartered

1.10

phones

1.09

 Head

1.03

quarters

1.03

quarter

1.03

ache

1.02

Head

0.96

lining

0.94

canon

0.94

butt

0.94

Activations Density 0.005%

proper nouns related to roles, names, and titles

references to leadership or authoritative figures

No Comments

No Known Activations