INDEX

Explanations

numeric values

oai_token-act-pair · gpt-3.5-turbo

sequences of characters or symbols that may not correspond to meaningful words in English or structured data

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

nesota

-0.96

merce

-0.94

puter

-0.94

ufact

-0.82

ittee

-0.81

istically

-0.81

wagen

-0.80

aido

-0.78

keepers

-0.77

ittees

-0.77

POSITIVE LOGITS

³

0.95

´

0.92

ł

0.88

een

0.84

ï¸ı

0.81

Ð»

0.81

Øª

0.80

Ð¾Ð

0.77

eed

0.74

ÑĢ

0.74

Activations Density 0.008%

numeric values

sequences of characters or symbols that may not correspond to meaningful words in English or structured data

No Comments

No Known Activations