INDEX

Explanations

phrases related to the concept of a topic or issue being brought up or discussed

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate a occurrence or introduction of a subject or event

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

amen

-0.64

AAF

-0.60

oret

-0.59

 Know

-0.57

rued

-0.56

coord

-0.55

ript

-0.54

hemat

-0.54

avid

-0.52

aya

-0.52

POSITIVE LOGITS

 comes

3.13

 Comes

2.18

 arrives

1.95

 arises

1.91

 came

1.91

comes

1.79

 emerges

1.76

 come

1.65

 goes

1.65

 derives

1.63

Activations Density 0.030%

phrases related to the concept of a topic or issue being brought up or discussed

phrases that indicate a occurrence or introduction of a subject or event

No Comments

No Known Activations