INDEX

Explanations

numbers and percentages followed by related words or phrases

oai_token-act-pair · gpt-3.5-turbo

quantities and statistical values related to data or metrics

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

ocracy

-0.64

imen

-0.61

gery

-0.60

 organisation

-0.59

activate

-0.57

organ

-0.57

 bureaucr

-0.57

rame

-0.56

uay

-0.56

craft

-0.55

POSITIVE LOGITS

 consecutive

0.98

th

0.96

ÃĹ

0.90

0.87

%-

0.82

Â½

0.82

 percent

0.82

âĢ²

0.81

 degrees

0.81

0.80

Activations Density 0.126%

numbers and percentages followed by related words or phrases

quantities and statistical values related to data or metrics

No Comments

No Known Activations