INDEX

Explanations

proper nouns or specific names related to news articles

oai_token-act-pair · gpt-3.5-turbo

references to files or documents

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ische

-0.76

itational

-0.75

orically

-0.74

utra

-0.74

atform

-0.72

psey

-0.70

ĵĺ

-0.70

orical

-0.69

osate

-0.68

erald

-0.68

POSITIVE LOGITS

Upload

0.87

 File

0.84

 FILE

0.83

 Photographer

0.79

upload

0.79

FILE

0.78

ette

0.74

 PHOTO

0.72

 Cabin

0.72

Path

0.71

Activations Density 0.012%

proper nouns or specific names related to news articles

references to files or documents

No Comments

No Known Activations