INDEX

Explanations

instances where an event or action is described as happening for the first or second time

oai_token-act-pair · gpt-3.5-turbo

instances of phrases indicating a repetition or occurrence for the first or second time

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

................................................................

-0.70

 accordingly

-0.69

 Mechdragon

-0.67

 meantime

-0.64

redients

-0.64

 Brilliant

-0.63

stellar

-0.62

 srfAttach

-0.61

çī

-0.60

Dat

-0.59

POSITIVE LOGITS

ever

0.83

 ever

0.81

 remotely

0.76

ibo

0.74

bered

0.70

 EVER

0.69

APH

0.69

 interacted

0.66

 bothered

0.65

 attempted

0.65

Activations Density 0.230%

instances where an event or action is described as happening for the first or second time

instances of phrases indicating a repetition or occurrence for the first or second time

No Comments

No Known Activations