INDEX

Explanations

phrases or words pertaining to specific locations

oai_token-act-pair · gpt-3.5-turbo

the specific term "vi" and its variations, possibly indicating a focus on visual elements or representations

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

*/(

-0.86

 Cosponsors

-0.74

lins

-0.66

spin

-0.65

raising

-0.65

 Kasich

-0.64

bread

-0.63

Ĥª

-0.63

ĭ

-0.62

rules

-0.61

POSITIVE LOGITS

Ã±

0.95

Äĩ

0.93

Äį

0.90

irus

0.87

uld

0.87

vi

0.86

ault

0.86

itri

0.82

plets

0.82

udence

0.81

Activations Density 0.006%

phrases or words pertaining to specific locations

the specific term "vi" and its variations, possibly indicating a focus on visual elements or representations

No Comments

No Known Activations