INDEX

Explanations

words related to awards, recognition, or superlatives

oai_token-act-pair · gpt-3.5-turbo

references to award categories and winners

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

IGHTS

-0.82

 inval

-0.68

aucuses

-0.67

tics

-0.66

idon

-0.63

umat

-0.61

 Greenpeace

-0.60

 falsely

-0.60

gypt

-0.60

 impatient

-0.59

POSITIVE LOGITS

seller

1.29

iary

1.11

selling

1.09

sell

1.01

 Practices

0.99

iaries

0.96

 Selling

0.90

 Answer

0.87

 Seller

0.82

 Worst

0.80

Activations Density 0.034%

words related to awards, recognition, or superlatives

references to award categories and winners

No Comments

No Known Activations