INDEX

Explanations

words related to statistics and analysis

oai_token-act-pair · gpt-3.5-turbo

conversational phrases and expressions of uncertainty or opinion

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Â®

-0.65

 Amid

-0.60

 Alas

-0.57

 Bare

-0.55

Yet

-0.53

yet

-0.52

Fey

-0.51

 Scroll

-0.50

 Avoid

-0.50

 Shroud

-0.50

POSITIVE LOGITS

 gonna

0.81

.'"

0.80

 incent

0.79

)."

0.78

entimes

0.77

0.75

 mathemat

0.73

 fuckin

0.72

plet

0.70

 everybody

0.70

Activations Density 1.858%

words related to statistics and analysis

conversational phrases and expressions of uncertainty or opinion

No Comments

No Known Activations