INDEX

Explanations

names of individuals, likely related to news articles or publications

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names of people and organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Frozen

-0.67

 idle

-0.64

Shape

-0.63

netflix

-0.63

 polar

-0.59

 favour

-0.57

adesh

-0.56

tics

-0.54

Downloadha

-0.52

 Rahul

-0.52

POSITIVE LOGITS

ONSORED

0.71

enson

0.63

ãĤ¼ãĤ¦ãĤ¹

0.60

heny

0.59

Jr

0.59

mann

0.58

nel

0.58

ertodd

0.57

 Kills

0.57

eyes

0.57

Activations Density 0.437%

names of individuals, likely related to news articles or publications

proper nouns, particularly names of people and organizations

No Comments

No Known Activations