INDEX

Explanations

terms related to the word "ams" or similar variations

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "ams" or similar recurring phrases

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

 wart

-0.67

 resemb

-0.61

hold

-0.61

 danger

-0.60

 hazard

-0.59

 causation

-0.59

 demographics

-0.59

GROUND

-0.58

 fluid

-0.58

 reversal

-0.58

POSITIVE LOGITS

mith

1.02

hip

1.01

hell

1.00

sterdam

0.94

olver

0.94

heet

0.92

icum

0.92

oufl

0.92

creen

0.91

chool

0.90

Activations Density 0.021%

terms related to the word "ams" or similar variations

references to the concept of "ams" or similar recurring phrases

No Comments

No Known Activations