INDEX

Explanations

expressions related to roles and responsibilities

oai_token-act-pair · gpt-3.5-turbo

references to individual contributions or roles within a collective context

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

 Combine

-0.64

 Result

-0.62

\">

-0.61

furt

-0.61

inctions

-0.57

due

-0.57

 Topic

-0.57

 Supplement

-0.56

heric

-0.56

CENT

-0.56

POSITIVE LOGITS

 brethren

0.82

stretched

0.80

 selves

0.79

 counterparts

0.76

 mates

0.74

iest

0.71

 mate

0.71

mates

0.68

 buddies

0.66

tre

0.66

Activations Density 0.258%

expressions related to roles and responsibilities

references to individual contributions or roles within a collective context

No Comments

No Known Activations