INDEX

Explanations

phrases related to how individuals are treated

oai_token-act-pair · gpt-3.5-turbo

instances of the word "treats" and its contexts in various scenarios

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

allow

-0.78

ItemImage

-0.78

 radius

-0.68

legraph

-0.67

perate

-0.66

OWN

-0.66

iring

-0.66

ortal

-0.65

 stroke

-0.64

opic

-0.64

POSITIVE LOGITS

 surprises

2.66

 treats

2.55

 goodies

1.95

 snacks

1.93

 prizes

1.75

 desserts

1.67

 sweets

1.47

 treasures

1.41

 tunes

1.41

 scraps

1.35

Activations Density 0.032%

phrases related to how individuals are treated

instances of the word "treats" and its contexts in various scenarios

No Comments

No Known Activations