INDEX

Explanations

terms related to notes or instructions

oai_token-act-pair · gpt-3.5-turbo

sentences or phrases that contain notes or comments

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

onet

-0.72

fu

-0.70

rail

-0.69

 lungs

-0.67

lled

-0.65

arters

-0.64

ãĤ¼ãĤ¦ãĤ¹

-0.62

 hinge

-0.61

stream

-0.60

REAM

-0.59

POSITIVE LOGITS

 Unsure

0.72

 Aires

0.72

 Beware

0.72

 Remastered

0.69

isks

0.65

 beware

0.64

Due

0.64

 NOTICE

0.64

 Always

0.63

 Remember

0.62

Activations Density 0.090%

terms related to notes or instructions

sentences or phrases that contain notes or comments

No Comments

No Known Activations