INDEX

Explanations

numbers with high activations occurring in a numeric order or sequence

oai_token-act-pair · gpt-3.5-turbo

numeric identifiers or ratings associated with entities

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

holder

-0.79

istically

-0.71

form

-0.68

think

-0.68

leck

-0.67

owicz

-0.66

 snipp

-0.66

ging

-0.66

estine

-0.65

 Sands

-0.64

POSITIVE LOGITS

mph

0.98

 ILCS

0.95

dB

0.90

0.86

rup

0.82

0.81

dB

0.81

0.78

é¾

0.77

Activations Density 0.023%

numbers with high activations occurring in a numeric order or sequence

numeric identifiers or ratings associated with entities

No Comments

No Known Activations