INDEX

Explanations

verbs followed by the word "to" indicating an action or intention

oai_token-act-pair · gpt-3.5-turbo

phrases indicating desires or intentions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Orders

-0.71

 Lines

-0.70

 Measures

-0.62

iple

-0.62

 Directions

-0.61

enforcement

-0.61

requires

-0.61

cases

-0.60

quickShipAvailable

-0.59

words

-0.59

POSITIVE LOGITS

 stay

1.08

be

1.06

 keep

1.00

 give

0.93

 emulate

0.89

 participate

0.88

 take

0.88

 make

0.87

 avoid

0.86

 perform

0.86

Activations Density 0.117%

verbs followed by the word "to" indicating an action or intention

phrases indicating desires or intentions

No Comments

No Known Activations