INDEX

Explanations

non-English characters that are not typically used in the English language

oai_token-act-pair · gpt-3.5-turbo

instances of significant events or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 eleph

-0.72

 tremend

-0.71

 metic

-0.69

 occas

-0.68

oun

-0.67

senal

-0.66

 undermin

-0.65

 unnecess

-0.64

 helicop

-0.63

ò

-0.63

POSITIVE LOGITS

Scroll

0.88

thinkable

0.70

%%%%

0.60

sr

0.59

initialized

0.59

Shadow

0.59

 isEnabled

0.58

acci

0.58

iola

0.58

\":

0.57

Activations Density 0.237%

non-English characters that are not typically used in the English language

instances of significant events or actions

No Comments

No Known Activations