INDEX

Explanations

proper nouns or names embedded in statements

oai_token-act-pair · gpt-3.5-turbo

statements of opinion or commentary on various topics

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

 actionGroup

-0.79

MpServer

-0.76

artifacts

-0.74

*/

-0.72

*/

-0.71

 warr

-0.70

icter

-0.70

FTWARE

-0.69

 Afgh

-0.68

corruption

-0.67

POSITIVE LOGITS

 quotes

1.05

 reply

1.04

 referring

0.98

 replied

0.97

 quotation

0.96

 quote

0.94

 replies

0.94

 echoed

0.94

yrics

0.92

 quoting

0.91

Activations Density 0.666%

proper nouns or names embedded in statements

statements of opinion or commentary on various topics

No Comments

No Known Activations