INDEX

Explanations

proper nouns or names

oai_token-act-pair · gpt-3.5-turbo

names or words relevant to specific individuals or characters

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

olutions

-0.82

acious

-0.81

ittal

-0.74

acies

-0.74

erion

-0.74

isco

-0.73

itals

-0.73

ikuman

-0.73

uers

-0.71

aldo

-0.70

POSITIVE LOGITS

hee

0.92

gee

0.91

 Springs

0.81

selage

0.80

 Kong

0.78

EEE

0.74

;;;;;;;;;;;;

0.72

 Choi

0.69

à©

0.68

HAHAHAHA

0.67

Activations Density 0.038%

proper nouns or names

names or words relevant to specific individuals or characters

No Comments

No Known Activations