INDEX

Explanations

phrases related to exceeding or surpassing something in performance or comparison

oai_token-act-pair · gpt-3.5-turbo

terms that indicate exceeding or outperforming

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

 Refuge

-0.68

 Origin

-0.66

Accessory

-0.65

 ACTION

-0.65

 Horror

-0.65

 Geneva

-0.65

Interstitial

-0.65

Downloadha

-0.62

 Goods

-0.61

Geo

-0.61

POSITIVE LOGITS

mented

1.10

uple

0.90

ighed

0.86

ceed

0.85

İĭ

0.83

Ķ

0.83

lex

0.82

rolet

0.82

pling

0.79

pload

0.78

Activations Density 0.015%

phrases related to exceeding or surpassing something in performance or comparison

terms that indicate exceeding or outperforming

No Comments

No Known Activations