INDEX

Explanations

specific references to the word "chains"

oai_token-act-pair · gpt-3.5-turbo

references to "chains," particularly in various contexts related to music, businesses, or entertainment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Gors

-0.76

sburgh

-0.74

dL

-0.73

issance

-0.73

 Niet

-0.71

 Galile

-0.69

sburg

-0.69

 Leah

-0.67

 Hague

-0.67

00200000

-0.66

POSITIVE LOGITS

 chains

1.13

aws

1.01

becue

0.87

layer

0.87

awed

0.85

ilver

0.85

chains

0.84

 chain

0.83

worms

0.81

ammy

0.81

Activations Density 0.004%

specific references to the word "chains"

references to "chains," particularly in various contexts related to music, businesses, or entertainment

No Comments

No Known Activations