INDEX

Explanations

terms related to data structuring or coding structures

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "structure" and its variations

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

 Greenwich

-0.80

hood

-0.73

 Clarkson

-0.73

 Leone

-0.70

 Sierra

-0.69

 Yellowstone

-0.67

cale

-0.64

 Holmes

-0.64

 Derby

-0.64

 proclamation

-0.63

POSITIVE LOGITS

urally

1.60

ured

1.17

ural

1.13

itionally

1.10

Struct

1.07

urer

1.03

uring

1.03

urers

1.02

struct

1.02

untarily

1.01

Activations Density 0.011%

terms related to data structuring or coding structures

occurrences of the word "structure" and its variations

No Comments

No Known Activations