INDEX

Explanations

information related to news articles or headlines

oai_token-act-pair · gpt-3.5-turbo

sentences that indicate completion or end points

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

©¶æ

-0.87

tymology

-0.77

 blacklist

-0.77

pse

-0.76

foundland

-0.73

 rebell

-0.71

 mascul

-0.69

 phylogen

-0.69

ãĥ¢

-0.68

idious

-0.68

POSITIVE LOGITS

 VIDEOS

1.10

 Buzz

1.02

 Video

0.93

 Caption

0.91

 TODAY

0.91

USA

0.87

CNN

0.86

 Time

0.85

CLOSE

0.81

 Provided

0.80

Activations Density 0.048%

information related to news articles or headlines

sentences that indicate completion or end points

No Comments

No Known Activations