INDEX

Explanations

phrases related to the number "two" or numerical pairs

oai_token-act-pair · gpt-3.5-turbo

references to a "two" concept, often related to timeframes or measurements

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

yles

-0.72

terday

-0.65

ivals

-0.63

rencies

-0.61

atoes

-0.61

ORTS

-0.60

 Worldwide

-0.59

raints

-0.58

 Rica

-0.58

POSITIVE LOGITS

fold

1.06

quel

0.96

teenth

0.90

 dimensional

0.83

0.81

acious

0.78

plus

0.77

peat

0.76

edged

0.76

dimensional

0.76

Activations Density 0.075%

phrases related to the number "two" or numerical pairs

references to a "two" concept, often related to timeframes or measurements

No Comments

No Known Activations