INDEX

Explanations

names of people or characters

oai_token-act-pair · gpt-3.5-turbo

phrases related to personal experiences and emotions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

requires

-0.73

 governs

-0.71

ierra

-0.68

onto

-0.64

 consumes

-0.64

ologies

-0.63

 assumes

-0.61

 improves

-0.61

ILA

-0.60

antage

-0.60

POSITIVE LOGITS

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

0.83

 happiest

0.81

urated

0.72

 last

0.68

KGB

0.67

 conceived

0.66

 instrumental

0.65

 inspiration

0.64

 revelation

0.64

itted

0.62

Activations Density 1.160%

names of people or characters

phrases related to personal experiences and emotions

No Comments

No Known Activations