INDEX

Explanations

general statements or phrases used to introduce broad ideas or concepts

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "generally" indicating general statements or observations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ÄŁ

-0.88

gur

-0.80

 Billion

-0.74

hao

-0.72

tein

-0.72

him

-0.71

lyn

-0.69

vana

-0.68

lez

-0.67

 Agenda

-0.67

POSITIVE LOGITS

 speaking

0.96

 regarded

0.94

 frowned

0.91

ensical

0.89

 exha

0.86

entimes

0.85

 behaved

0.84

 accepted

0.84

 identifiable

0.83

 assumed

0.81

Activations Density 0.011%

general statements or phrases used to introduce broad ideas or concepts

occurrences of the word "generally" indicating general statements or observations

No Comments

No Known Activations

general statements or phrases used to introduce broad ideas or concepts

occurrences of the word "generally" indicating general statements or observations

No Comments

No Known Activations