INDEX

Explanations

instances where something is described positively with words like 'well.'

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate successful performance or positive outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

inity

-0.76

rush

-0.76

hip

-0.76

illary

-0.71

gy

-0.68

IENCE

-0.65

 Samp

-0.63

ory

-0.61

hyde

-0.60

erick

-0.59

POSITIVE LOGITS

enough

0.79

 suited

0.74

espie

0.72

NESS

0.71

 enough

0.70

lied

0.70

 liked

0.70

nesses

0.69

reads

0.69

 behaved

0.68

Activations Density 0.055%

instances where something is described positively with words like 'well.'

phrases that indicate successful performance or positive outcomes

No Comments

No Known Activations