INDEX

Explanations

mentions of financial terms and numbers

oai_token-act-pair · gpt-3.5-turbo

punctuations, particularly commas, indicating lists or clauses

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

iple

-0.92

ibo

-0.71

chin

-0.71

ety

-0.67

uit

-0.65

jured

-0.64

endo

-0.64

power

-0.64

reth

-0.63

UF

-0.63

POSITIVE LOGITS

 namely

1.10

 albeit

1.06

 lest

0.95

 culminating

0.94

 although

0.92

 whereby

0.89

 which

0.88

 wherein

0.85

 whereas

0.85

 implying

0.83

Activations Density 1.086%

mentions of financial terms and numbers

punctuations, particularly commas, indicating lists or clauses

No Comments

No Known Activations