INDEX

Explanations

words related to connections or links between different entities

oai_token-act-pair · gpt-3.5-turbo

references to connections or associations between different entities or concepts

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

nesty

-0.69

olan

-0.68

 plum

-0.68

ople

-0.66

opia

-0.64

conserv

-0.63

sein

-0.63

arten

-0.62

draft

-0.62

actic

-0.61

POSITIVE LOGITS

 connected

3.54

connected

2.82

 interconnected

2.07

 disconnected

1.93

 linked

1.83

 connect

1.83

 connects

1.80

 connecting

1.77

 Connect

1.68

 connection

1.63

Activations Density 0.013%

words related to connections or links between different entities

references to connections or associations between different entities or concepts

No Comments

No Known Activations