INDEX

Explanations

verbs related to investigating or researching

oai_token-act-pair · gpt-3.5-turbo

instances of the word "explore."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

enforcement

-0.71

corn

-0.70

appointed

-0.67

roll

-0.67

invoke

-0.67

enta

-0.66

warning

-0.66

gage

-0.65

cakes

-0.64

remember

-0.64

POSITIVE LOGITS

 avenues

1.02

 possibilities

1.00

 deeper

0.96

 anew

0.92

ationally

0.91

 further

0.89

 topics

0.86

 unexpl

0.84

 ways

0.84

 themes

0.84

Activations Density 0.066%

verbs related to investigating or researching

instances of the word "explore."

No Comments

No Known Activations