INDEX

Explanations

questions or statements followed by commas

oai_token-act-pair · gpt-3.5-turbo

punctuation and specific interrogative expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 inval

-0.77

ixel

-0.67

://

-0.67

 treasury

-0.65

 gobl

-0.64

 Dealer

-0.63

 desper

-0.62

 murderers

-0.59

 corrid

-0.58

 condem

-0.58

POSITIVE LOGITS

uh

0.71

then

0.69

say

0.67

eenth

0.67

Sund

0.65

um

0.62

 Amen

0.61

usually

0.61

orians

0.60

 pray

0.60

Activations Density 0.163%

questions or statements followed by commas

punctuation and specific interrogative expressions

No Comments

No Known Activations