INDEX

Explanations

references to specific geographical locations, particularly related to political conflicts

oai_token-act-pair · gpt-3.5-turbo

references to the Gaza Strip and related territorial terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥ¼ãĥĨ

-0.85

wcsstore

-0.80

ãĤ¤ãĥĪ

-0.79

Lear

-0.78

ãĥīãĥ©

-0.75

istically

-0.71

ãĤ¢ãĥ«

-0.71

fig

-0.69

é¾įå

-0.69

 obser

-0.69

POSITIVE LOGITS

 Territories

0.99

istan

0.78

 Territory

0.77

 annexed

0.74

 ravaged

0.74

 Affairs

0.73

 Sudan

0.71

 Strip

0.71

 Hezbollah

0.69

 Quart

0.69

Activations Density 0.037%

references to specific geographical locations, particularly related to political conflicts

references to the Gaza Strip and related territorial terms

No Comments

No Known Activations