INDEX

Explanations

dates or events described as happening historically

oai_token-act-pair · gpt-3.5-turbo

references to historical information or 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

MO

-0.69

reen

-0.66

 Sunshine

-0.66

wra

-0.64

 Bunny

-0.64

 Toro

-0.63

 sunglasses

-0.63

wrapper

-0.63

 Caller

-0.63

ared

-0.61

POSITIVE LOGITS

Hist

3.53

 Hist

2.70

hist

2.34

 Historical

1.93

 hist

1.76

 historians

1.51

 Historic

1.44

 historian

1.39

History

1.32

 History

1.25

Activations Density 0.029%

dates or events described as happening historically

references to historical information or contexts

No Comments

No Known Activations