INDEX

Explanations

single words or short phrases within a larger context

oai_token-act-pair · gpt-3.5-turbo

mentions of lists or enumerations

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

mentions of "list" in various contexts.

oai_token-act-pair · gpt-4o Triggered by @tatsatx

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Aber

-0.75

 Pagan

-0.62

 stride

-0.61

 Advocate

-0.61

 Huck

-0.60

 merry

-0.59

 Gore

-0.57

 reper

-0.57

 Farmer

-0.56

 Advantage

-0.56

POSITIVE LOGITS

erv

1.18

ening

1.03

icle

0.96

icles

0.95

ener

0.94

eners

0.93

abet

0.86

enable

0.84

lists

0.83

erve

0.83

Activations Density 0.066%

single words or short phrases within a larger context

mentions of lists or enumerations

mentions of "list" in various contexts.

No Comments

No Known Activations