INDEX

Explanations

names of individuals

oai_token-act-pair · gpt-3.5-turbo

conjunctions and phrases indicating a connection between subjects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

girls

-0.84

Characters

-0.82

days

-0.80

pees

-0.80

isks

-0.79

Players

-0.76

NetMessage

-0.74

different

-0.74

yles

-0.74

issions

-0.74

POSITIVE LOGITS

 longtime

1.22

 former

1.15

 whistleblower

1.12

 colleague

1.11

 collaborator

1.11

 outspoken

1.09

 ardent

1.08

 educator

1.07

 activist

1.07

 fellow

1.07

Activations Density 0.077%

names of individuals

conjunctions and phrases indicating a connection between subjects

No Comments

No Known Activations