INDEX

Explanations

instances of the word "flags"

oai_token-act-pair · gpt-3.5-turbo

references to flags, particularly in political or national contexts

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

liv

-0.78

abolic

-0.78

achment

-0.74

vana

-0.72

hya

-0.71

pez

-0.71

Sco

-0.71

Yel

-0.70

 Ther

-0.69

 Rehab

-0.68

POSITIVE LOGITS

 flags

3.97

 Flags

2.78

 flag

2.57

flags

2.51

Flags

2.40

flag

1.96

Flag

1.93

 Flag

1.87

 banners

1.76

FLAG

1.49

Activations Density 0.016%

instances of the word "flags"

references to flags, particularly in political or national contexts

No Comments

No Known Activations