INDEX

Explanations

mention of images or pictures

oai_token-act-pair · gpt-3.5-turbo

references to pictures or images

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

NetMessage

-0.81

avorite

-0.69

interstitial

-0.69

 trave

-0.65

edient

-0.63

ifted

-0.61

 concentrate

-0.61

rane

-0.60

 footing

-0.60

 shroud

-0.59

POSITIVE LOGITS

colo

1.16

pic

1.06

 pics

0.91

wikipedia

0.85

 https

0.83

pic

0.82

chrom

0.78

Pic

0.76

DOI

0.76

https

0.76

Activations Density 0.008%

mention of images or pictures

references to pictures or images

No Comments

No Known Activations