INDEX

Explanations

profane and emphatic expressions

oai_token-act-pair · gpt-3.5-turbo

expressions of frustration or strong emphasis

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

IDS

-0.69

OHN

-0.68

DN

-0.64

KEN

-0.63

 Flavoring

-0.62

quo

-0.61

ynthesis

-0.60

KY

-0.57

amide

-0.57

":""},{"

-0.57

POSITIVE LOGITS

near

1.05

ibly

1.01

ation

0.86

atio

0.84

orse

0.77

 near

0.74

ably

0.72

ated

0.72

 damned

0.72

emed

0.72

Activations Density 0.024%

profane and emphatic expressions

expressions of frustration or strong emphasis

No Comments

No Known Activations