INDEX

Explanations

joint words or phrases including "hem"

oai_token-act-pair · gpt-3.5-turbo

references to the word "hem."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ULTS

-0.72

âķĲ

-0.72

çĶŁ

-0.70

raught

-0.68

AVG

-0.67

 Peaks

-0.66

 ranking

-0.64

 Puerto

-0.61

rieving

-0.61

ŃĶ

-0.61

POSITIVE LOGITS

isphere

1.12

oglobin

1.03

oths

1.00

icals

0.98

ophon

0.95

selves

0.93

hem

0.91

med

0.89

iland

0.89

ming

0.86

Activations Density 0.007%

joint words or phrases including "hem"

references to the word "hem."

No Comments

No Known Activations