INDEX

Explanations

short phrases related to specific events or situations

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, particularly periods and commas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 affili

-0.91

 appropriate

-0.75

 criminal

-0.75

 involuntary

-0.73

 unle

-0.72

 temperament

-0.72

 commodities

-0.72

 sustainable

-0.71

 heirs

-0.70

 attain

-0.69

POSITIVE LOGITS

 Afterwards

1.43

 Didn

1.32

 Unfortunately

1.26

 Seems

1.24

 Basically

1.21

 Turns

1.20

 Apparently

1.19

 Needless

1.19

 Luckily

1.18

 Then

1.14

Activations Density 0.494%

short phrases related to specific events or situations

punctuation marks, particularly periods and commas

No Comments

No Known Activations