INDEX

Explanations

words related to concepts or actions that involve decision-making or contrasting choices

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 WATCHED

-0.63

soon

-0.63

 whereas

-0.63

âĹ¼

-0.62

 Authorization

-0.59

estern

-0.59

mma

-0.58

 actionGroup

-0.57

 Judges

-0.56

 Nanto

-0.55

POSITIVE LOGITS

 necessarily

0.84

 ones

0.84

nor

0.83

 altogether

0.77

 itself

0.76

 outright

0.76

 anymore

0.69

 actual

0.67

olves

0.64

 oneself

0.64

Activations Density 0.639%

words related to concepts or actions that involve decision-making or contrasting choices

No Comments

No Known Activations