INDEX

Explanations

numerical values indicating a quantity or a percentage followed by a specific unit of measurement or term

oai_token-act-pair · gpt-3.5-turbo

quantitative data related to thresholds and limits

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

 ãĤµ

-0.67

 Fallen

-0.66

 Truman

-0.64

'[

-0.60

ornia

-0.60

PLA

-0.59

 dearly

-0.58

Bal

-0.58

 Brav

-0.58

 valiant

-0.57

POSITIVE LOGITS

 percent

0.81

 anymore

0.79

(<

0.78

GHz

0.72

%.

0.72

 cents

0.68

 diagnose

0.65

escent

0.65

%;

0.65

utenberg

0.64

Activations Density 0.182%

numerical values indicating a quantity or a percentage followed by a specific unit of measurement or term

quantitative data related to thresholds and limits

No Comments

No Known Activations