52 lines
1.4 KiB
Markdown
52 lines
1.4 KiB
Markdown
---
|
|
language:
|
|
- en
|
|
tags:
|
|
- causal-lm
|
|
- llama
|
|
- point-in-time
|
|
- dated
|
|
- lookahead-bias-free
|
|
pipeline_tag: text-generation
|
|
---
|
|
|
|
# DatedGPT-2016 (base)
|
|
|
|
**DatedGPT** is a family of point-in-time language models: each vintage is
|
|
trained only on data available up to its cutoff date, making it suitable for
|
|
lookahead-bias-free prediction and point-in-time analysis.
|
|
|
|
This is the **base (pretrained) model** with data up to **2016** — no
|
|
instruction tuning. For the instruction-tuned variants, see the
|
|
`datedgpt-instruct-*` repositories in this organization.
|
|
|
|
| Property | Value |
|
|
|----------|-------|
|
|
| Architecture | LlamaForCausalLM |
|
|
| Parameters | ~1.3 B |
|
|
| Context length | 2048 |
|
|
| Vocab | 32,000 (SentencePiece) |
|
|
| Precision | bfloat16 |
|
|
| Data vintage | 2016 |
|
|
|
|
## Usage
|
|
|
|
```python
|
|
import torch
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
|
|
repo_id = "datedgpt/datedgpt-2016-base"
|
|
tokenizer = AutoTokenizer.from_pretrained(repo_id)
|
|
model = AutoModelForCausalLM.from_pretrained(repo_id, torch_dtype=torch.bfloat16, device_map="auto")
|
|
|
|
inputs = tokenizer("The stock market in 2016", return_tensors="pt").to(model.device)
|
|
output = model.generate(**inputs, max_new_tokens=64, use_cache=True)
|
|
print(tokenizer.decode(output[0], skip_special_tokens=True))
|
|
```
|
|
|
|
## Limitations
|
|
|
|
- Base model: completions only, no chat/instruction following.
|
|
- Knowledge limited to the 2016 data vintage.
|
|
- No RLHF or safety tuning.
|