INDEX

Explanations

phrases indicating ease or simplicity

oai_token-act-pair · gpt-3.5-turbo

phrases that suggest ease in completing actions or making assessments

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Dur

-0.90

ELD

-0.79

cedented

-0.72

Strong

-0.68

Room

-0.68

 TAMADRA

-0.66

 Relief

-0.66

Choice

-0.66

 Brach

-0.65

soever

-0.64

POSITIVE LOGITS

 forget

1.24

 visualize

1.17

 miscon

1.15

 confuse

1.15

 imagine

1.12

 overlook

1.12

 convince

1.11

 identify

1.07

 dismiss

1.06

 envision

1.06

Activations Density 0.066%

phrases indicating ease or simplicity

phrases that suggest ease in completing actions or making assessments

No Comments

No Known Activations