INDEX

Explanations

names or terms related to specified individuals or projects

oai_token-act-pair · gpt-3.5-turbo

instances of a specific name and its variations, especially related to legal or news contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

vill

-0.69

 Witch

-0.68

venge

-0.68

agg

-0.67

 Wolves

-0.66

Spo

-0.65

war

-0.65

glass

-0.64

Sor

-0.64

War

-0.63

POSITIVE LOGITS

ERT

4.03

redo

1.44

 Bold

1.44

 flagged

1.05

ERG

1.03

OTS

1.02

RSA

0.99

ert

0.91

ERN

0.88

IRED

0.87

Activations Density 0.021%

names or terms related to specified individuals or projects

instances of a specific name and its variations, especially related to legal or news contexts

No Comments

No Known Activations