INDEX

Explanations

phrases indicating a high level of importance or value to something

oai_token-act-pair · gpt-3.5-turbo

phrases indicating something significant is at risk or in jeopardy

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

nces

-0.76

selves

-0.71

marine

-0.67

ments

-0.64

comed

-0.62

cert

-0.62

period

-0.62

ĺ

-0.61

 conditioned

-0.60

bite

-0.59

POSITIVE LOGITS

 stake

1.43

onement

1.08

 hand

1.00

 least

0.93

 heart

0.86

 risk

0.84

yp

0.83

 odds

0.82

abase

0.81

 play

0.79

Activations Density 0.071%

phrases indicating a high level of importance or value to something

phrases indicating something significant is at risk or in jeopardy

No Comments

No Known Activations