INDEX

Explanations

specific nouns and terms related to news articles or reports

oai_token-act-pair · gpt-3.5-turbo

references to specific attacks, reports, or events involving the term "the" in various contexts

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

isphere

-0.86

thia

-0.70

 distingu

-0.68

Ń·

-0.68

 âĢº

-0.67

dor

-0.66

ovember

-0.65

umbered

-0.65

puted

-0.64

cliffe

-0.63

POSITIVE LOGITS

 bluff

1.25

 shots

0.95

 kettle

0.74

 hotline

0.72

 Shots

0.69

 situation

0.63

 actions

0.63

 attention

0.62

act

0.62

 halt

0.61

Activations Density 0.094%

specific nouns and terms related to news articles or reports

references to specific attacks, reports, or events involving the term "the" in various contexts

No Comments

No Known Activations