INDEX

Explanations

positions or roles held by individuals in different organizations or fields

oai_token-act-pair · gpt-3.5-turbo

contexts where the word "as" introduces roles or identifications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

yrs

-0.77

 VIDEOS

-0.76

ramid

-0.72

oes

-0.70

ople

-0.69

itiveness

-0.68

aceae

-0.65

tons

-0.64

roying

-0.64

itational

-0.62

POSITIVE LOGITS

pired

1.07

pires

1.01

 well

1.00

ynchron

0.92

 opposed

0.89

 ambassador

0.85

bestos

0.85

 deputy

0.84

 trustee

0.84

 assistant

0.82

Activations Density 0.112%

positions or roles held by individuals in different organizations or fields

contexts where the word "as" introduces roles or identifications

No Comments

No Known Activations