INDEX

Explanations

warnings or alerts signaling potential issues or dangers

oai_token-act-pair · gpt-3.5-turbo

references to warnings

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

RG

-0.69

olved

-0.67

entity

-0.67

pole

-0.66

iga

-0.65

ashion

-0.65

 ÃĹ

-0.65

 âĸ

-0.64

ovo

-0.64

ater

-0.64

POSITIVE LOGITS

 warnings

3.97

 warning

2.28

 Warn

2.01

warn

1.88

 warn

1.75

 advis

1.74

warning

1.73

 alerts

1.71

 Warning

1.71

 warns

1.63

Activations Density 0.021%

warnings or alerts signaling potential issues or dangers

references to warnings

No Comments

No Known Activations