INDEX

Explanations

phrases related to discussions, opinions, evaluations, and reactions

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, particularly commas, within a text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ãĤ´ãĥ³

-1.03

è£ħ

-0.87

çīĪ

-0.75

ãģ®å

-0.74

ãĥ¯

-0.74

éŃĶ

-0.73

vre

-0.70

%:

-0.70

OVA

-0.68

ãĤ¨ãĥ«

-0.67

POSITIVE LOGITS

 yeah

1.20

 whereas

1.08

 because

1.01

 obviously

0.99

 frankly

0.99

but

0.98

 blah

0.96

 which

0.95

0.94

 okay

0.94

Activations Density 0.286%

phrases related to discussions, opinions, evaluations, and reactions

punctuation marks, particularly commas, within a text

No Comments

No Known Activations