INDEX

Explanations

words indicating movement or direction

oai_token-act-pair · gpt-3.5-turbo

expressions that indicate direction or progression

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

nz

-0.78

CD

-0.76

RC

-0.74

NZ

-0.70

cell

-0.69

cham

-0.69

umm

-0.67

named

-0.67

nai

-0.65

drivers

-0.65

POSITIVE LOGITS

 toward

1.29

 towards

1.09

ward

0.90

vernment

0.83

 Towards

0.81

 infinity

0.80

outheast

0.78

WARD

0.78

wards

0.78

 fruition

0.74

Activations Density 0.011%

words indicating movement or direction

expressions that indicate direction or progression

No Comments

No Known Activations