INDEX

Explanations

words related to a specific geographical location, possibly a country or region

oai_token-act-pair · gpt-3.5-turbo

references to Transnistria

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

 whit

-0.75

 Presidents

-0.71

 oppos

-0.65

 tampering

-0.62

 overcl

-0.62

 Tough

-0.62

 exped

-0.61

 elect

-0.61

 ageing

-0.59

 acqu

-0.58

POSITIVE LOGITS

ria

4.70

rian

2.40

rians

2.22

rium

1.62

rius

1.57

rio

1.52

riel

1.50

ri

1.40

rien

1.38

rial

1.35

Activations Density 0.008%

words related to a specific geographical location, possibly a country or region

references to Transnistria

No Comments

No Known Activations