INDEX

Explanations

proper nouns and names of people or organizations

oai_token-act-pair · gpt-3.5-turbo

references to award nominations and winners

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

 humidity

-0.69

unct

-0.66

 engulf

-0.63

 trenches

-0.62

 disturbance

-0.61

 overloaded

-0.61

 grinding

-0.57

 surface

-0.57

amination

-0.57

 haze

-0.56

POSITIVE LOGITS

ĳå£«

0.69

ername

0.67

 waivers

0.66

 Congratulations

0.64

$$$$

0.64

eligible

0.63

 Cosponsors

0.62

è£

0.60

çĦ

0.60

 Cabinet

0.59

Activations Density 1.333%

proper nouns and names of people or organizations

references to award nominations and winners

No Comments

No Known Activations