INDEX

Explanations

sentences or phrases ending with punctuation marks typically indicating conclusion or emphasis

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

isbury

-0.77

 applauded

-0.74

ŃĶ

-0.74

 honorable

-0.73

 supporter

-0.70

 believer

-0.70

 affili

-0.70

 compliment

-0.69

 indistinguishable

-0.69

irmed

-0.68

POSITIVE LOGITS

 Luckily

1.98

 Fortunately

1.94

 Worse

1.90

 Thankfully

1.82

 Therefore

1.73

 Consequently

1.72

 Nonetheless

1.57

 Nevertheless

1.56

Luckily

1.55

 Hence

1.54

Activations Density 0.499%

sentences or phrases ending with punctuation marks typically indicating conclusion or emphasis

No Comments

No Known Activations