INDEX

Explanations

phrases or words related to enabling or allowing something

oai_token-act-pair · gpt-3.5-turbo

phrases related to granting or providing capabilities

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

Coh

-0.68

 tart

-0.64

 Baker

-0.62

Sum

-0.61

Civ

-0.61

 Cold

-0.60

ians

-0.60

ARS

-0.60

 Scream

-0.59

itol

-0.58

POSITIVE LOGITS

 enable

3.30

 enabled

2.09

 enables

2.02

 Enable

1.96

 enabling

1.88

enable

1.87

Enable

1.86

 disable

1.72

 allow

1.61

enabled

1.49

Activations Density 0.022%

phrases or words related to enabling or allowing something

phrases related to granting or providing capabilities

No Comments

No Known Activations