INDEX

Explanations

the word "what" followed by either a question or a description of something

oai_token-act-pair · gpt-3.5-turbo

the word "what" in various contexts, suggesting an exploration of inquiry or questioning

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 nationwide

-0.61

 Mock

-0.58

 misinterpret

-0.56

uating

-0.55

enger

-0.54

enburg

-0.54

ointed

-0.53

 defect

-0.53

 Warn

-0.52

 eaves

-0.52

POSITIVE LOGITS

soever

1.63

 constitutes

1.01

 happens

0.97

 amounts

0.91

 else

0.90

 follows

0.85

 appears

0.81

 feels

0.80

?).

0.79

 seems

0.79

Activations Density 0.076%

the word "what" followed by either a question or a description of something

the word "what" in various contexts, suggesting an exploration of inquiry or questioning

No Comments

No Known Activations