INDEX

Explanations

technical terms and phrases related to information sources and industries

oai_token-act-pair · gpt-3.5-turbo

references to notable events or entities in pop culture or entertainment

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

 deduct

-0.72

 calculus

-0.72

 juveniles

-0.71

 reading

-0.70

 centrif

-0.70

 spit

-0.68

 pulp

-0.68

 freezer

-0.68

 corros

-0.68

 duct

-0.67

POSITIVE LOGITS

Tra

1.01

Tech

1.00

Hour

0.98

Town

0.97

Hon

0.95

Specific

0.95

Metal

0.94

Cent

0.94

Mail

0.94

Day

0.93

Activations Density 0.119%

technical terms and phrases related to information sources and industries

references to notable events or entities in pop culture or entertainment

No Comments

No Known Activations