INDEX

Explanations

words related to a particular individual named "Ghomeshi"

oai_token-act-pair · gpt-3.5-turbo

the repeated mentions of a specific name, indicating prominent figures in a narrative

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

DEP

-0.74

arching

-0.71

į

-0.70

irl

-0.67

sk

-0.66

STER

-0.64

 takeoff

-0.64

Æ

-0.63

 Fundamental

-0.63

EStreamFrame

-0.61

POSITIVE LOGITS

omes

1.27

ome

0.98

velt

0.90

chool

0.89

OME

0.79

itus

0.74

chu

0.73

ppa

0.72

agues

0.71

combe

0.71

Activations Density 0.009%

words related to a particular individual named "Ghomeshi"

the repeated mentions of a specific name, indicating prominent figures in a narrative

No Comments

No Known Activations