INDEX

Explanations

names or terms related to high-ranking positions, possibly in a political or organizational context

oai_token-act-pair · gpt-3.5-turbo

proper nouns and names, particularly associated with individuals and organizations

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

 resil

-0.92

EStream

-0.84

 darn

-0.77

glers

-0.74

cffff

-0.74

 glim

-0.73

 trough

-0.72

 tremend

-0.70

 mathemat

-0.69

 obser

-0.69

POSITIVE LOGITS

hyde

0.95

enko

0.93

 Gomez

0.87

opsis

0.82

 Gonzalez

0.80

aga

0.77

hea

0.77

quez

0.76

 Banks

0.75

anne

0.75

Activations Density 0.278%

names or terms related to high-ranking positions, possibly in a political or organizational context

proper nouns and names, particularly associated with individuals and organizations

No Comments

No Known Activations