INDEX

Explanations

phrases indicating strong emotional reactions or opinions

oai_token-act-pair · gpt-3.5-turbo

expressions of strong personal sentiment or emphasis

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

 guiActiveUnfocused

-0.80

adena

-0.66

 Presidency

-0.64

 Cutting

-0.63

AST

-0.61

 Organizations

-0.61

ESCO

-0.60

idding

-0.60

 Georgian

-0.60

olding

-0.59

POSITIVE LOGITS

owe

1.14

 knew

1.07

 forgot

0.97

 need

0.94

 tasted

0.92

 ought

0.92

 wrote

0.91

 despise

0.90

 manage

0.90

 swore

0.89

Activations Density 0.218%

phrases indicating strong emotional reactions or opinions

expressions of strong personal sentiment or emphasis

No Comments

No Known Activations