INDEX

Explanations

structural elements related to data or programming code

oai_token-act-pair · gpt-3.5-turbo

structured data or key-value pairs, likely in a programming context

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

 Mush

-0.79

issance

-0.68

 Heck

-0.68

 Guilty

-0.68

 Flynn

-0.66

 Cary

-0.65

 Yose

-0.64

 Fine

-0.63

 Smoke

-0.62

Cry

-0.61

POSITIVE LOGITS

tags

1.06

description

1.00

duration

1.00

price

0.97

published

0.95

height

0.94

},

0.93

soType

0.91

priority

0.91

language

0.90

Activations Density 0.052%

structural elements related to data or programming code

structured data or key-value pairs, likely in a programming context

No Comments

No Known Activations