INDEX

Explanations

phrases related to time durations and sequences

oai_token-act-pair · gpt-3.5-turbo

consecutive phrases that describe length and trying experiences

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

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

Alert

-0.72

selves

-0.70

trak

-0.69

 Meow

-0.68

foundland

-0.66

 Warfare

-0.65

Ö¼

-0.65

ģ«

-0.65

 Blueprint

-0.62

 Cosponsors

-0.62

POSITIVE LOGITS

 subsequ

0.89

duration

0.82

 ardu

0.77

(>

0.75

 longest

0.74

distance

0.70

 duration

0.70

 awaited

0.70

 distances

0.69

ago

0.69

Activations Density 0.167%

phrases related to time durations and sequences

consecutive phrases that describe length and trying experiences

No Comments

No Known Activations