INDEX

Explanations

references to the news agency AFP

oai_token-act-pair · gpt-3.5-turbo

mentions of the news agency AFP

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

stood

-0.92

sbm

-0.73

pron

-0.71

 Sears

-0.68

 Heavenly

-0.67

thal

-0.66

tons

-0.65

gage

-0.65

 Simpson

-0.64

comes

-0.64

POSITIVE LOGITS

UTERS

0.95

 PHOTO

0.85

sburg

0.85

DEF

0.81

enhagen

0.80

AFP

0.79

TN

0.76

aukee

0.76

DEP

0.73

attle

0.72

Activations Density 0.009%

references to the news agency AFP

mentions of the news agency AFP

No Comments

No Known Activations