INDEX

Explanations

prepositions followed by descriptions of various types of attacks

oai_token-act-pair · gpt-3.5-turbo

references to various attacks, particularly their occurrence and impact

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

uid

-0.95

drawn

-0.87

ãĤ´ãĥ³

-0.82

 wrinkles

-0.80

uber

-0.72

zl

-0.71

lied

-0.71

00200000

-0.70

vec

-0.70

grow

-0.68

POSITIVE LOGITS

 Pearl

0.91

 civilians

0.90

itals

0.90

 unarmed

0.88

 innocent

0.82

 embassies

0.80

 Tours

0.78

 unprotected

0.76

 Gaza

0.74

 civilian

0.74

Activations Density 0.213%

prepositions followed by descriptions of various types of attacks

references to various attacks, particularly their occurrence and impact

No Comments

No Known Activations