INDEX

Explanations

phrases related to superlatives or rankings, such as "best", "biggest", or "most"

oai_token-act-pair · gpt-3.5-turbo

phrases that denote significant rankings or listings of events, items, or achievements

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ô

-0.72

uyomi

-0.69

otom

-0.68

ISO

-0.66

oma

-0.65

OA

-0.64

MS

-0.64

KK

-0.63

hyde

-0.62

bid

-0.61

POSITIVE LOGITS

 theirs

1.15

 hers

0.98

 ours

0.93

 yours

0.82

course

0.78

 course

0.76

 sorts

0.75

 Week

0.73

il

0.72

 anywhere

0.71

Activations Density 0.100%

phrases related to superlatives or rankings, such as "best", "biggest", or "most"

phrases that denote significant rankings or listings of events, items, or achievements

No Comments

No Known Activations