INDEX

Explanations

numeric values related to levels, figures, or statistics

oai_token-act-pair · gpt-3.5-turbo

instances of financial metrics or statistics regarding their lowest or highest levels

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

OTA

-0.84

CHAT

-0.73

è»

-0.72

 alike

-0.69

AAF

-0.67

love

-0.67

bart

-0.64

å½

-0.62

Future

-0.60

âĹ¼

-0.59

POSITIVE LOGITS

ahime

0.89

flation

0.82

verages

0.80

 averages

0.76

vati

0.76

 inflation

0.75

ibaba

0.74

itures

0.70

avg

0.68

 projected

0.67

Activations Density 0.245%

numeric values related to levels, figures, or statistics

instances of financial metrics or statistics regarding their lowest or highest levels

No Comments

No Known Activations