INDEX

Explanations

phrases related to cancellations or interruptions

oai_token-act-pair · gpt-3.5-turbo

phrases related to cancellations or reasons for interruptions

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

FIELD

-0.82

enos

-0.76

ulse

-0.73

pton

-0.69

podcast

-0.68

upper

-0.65

PI

-0.65

REE

-0.65

clair

-0.65

soDeliveryDate

-0.64

POSITIVE LOGITS

 prematurely

0.88

 altogether

0.81

 fearing

0.78

 abruptly

0.77

livion

0.73

 permanently

0.72

 replaced

0.71

 deletion

0.71

 citing

0.71

 disgrace

0.71

Activations Density 0.885%

phrases related to cancellations or interruptions

phrases related to cancellations or reasons for interruptions

No Comments

No Known Activations