INDEX

Explanations

phrases related to news and information sharing

oai_token-act-pair · gpt-3.5-turbo

mentions of significant events or developments

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Blair

-0.82

 Everett

-0.80

 velvet

-0.79

 tornado

-0.78

Boo

-0.77

 Frankenstein

-0.77

 Winchester

-0.76

 Floyd

-0.75

 Vortex

-0.75

 Jacksonville

-0.74

POSITIVE LOGITS

According

1.65

Regarding

1.65

Meanwhile

1.63

Besides

1.62

Refer

1.61

Asked

1.61

Moreover

1.61

Earlier

1.59

Talking

1.58

However

1.57

Activations Density 0.305%

phrases related to news and information sharing

mentions of significant events or developments

No Comments

No Known Activations