INDEX

Explanations

names of specific entities, such as people, places, and organizations

oai_token-act-pair · gpt-3.5-turbo

names of media, locations, and brands

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

="#

-0.70

ËĪ

-0.68

-0.63

ileaks

-0.61

HAS

-0.61

ãĥ£

-0.60

ãĥ¼ãĥĨãĤ£

-0.59

endum

-0.58

":"/

-0.56

tains

-0.56

POSITIVE LOGITS

 respectively

1.78

 alike

1.12

Interstitial

0.93

+.

0.88

0.81

0.80

*.

0.80

 Aven

0.80

etc

0.77

 combined

0.76

Activations Density 0.411%

names of specific entities, such as people, places, and organizations

names of media, locations, and brands

No Comments

No Known Activations