Skip to content

Switch TinyLlama pretraining back to 16-mixed #882

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jan 15, 2024
Merged

Switch TinyLlama pretraining back to 16-mixed #882

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion pretrain/tinyllama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,7 +70,7 @@ def setup(resume: Union[bool, Path] = False):
logger = choose_logger(logger_name, name=name, resume=resume)

strategy = FSDPStrategy(auto_wrap_policy={Block}, state_dict_type="full", sharding_strategy="HYBRID_SHARD")
fabric = L.Fabric(devices=devices, strategy=strategy, precision="bf16-true", loggers=[logger])
fabric = L.Fabric(devices=devices, strategy=strategy, precision="bf16-mixed", loggers=[logger])
fabric.launch()

fabric.print(hparams)
Expand Down
23 changes: 12 additions & 11 deletions tutorials/pretrain_tinyllama.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,17 +11,18 @@ This tutorial will walk you through pretraining [TinyLlama](https://github.com/j

Here is a quick fact sheet:

| Name | Description |
|------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------------------------|
| Parameters | 1.1B |
| Model Size | Layers: 22, Heads: 32, Query Groups: 4, Embedding Size: 2048, Intermediate Size: 5632 |
| Sequence Length | 2048 |
| Batch Size | 2 million tokens (2048 * 1024) |
| Learning Rate | 4e-4 |
| Learning Rate Schedule | Cosine with 2000 warmup steps |
| Training Data | [SlimPajama](https://huggingface.co/datasets/cerebras/slimpajama-627b) (893 GB), [Starcoder](https://huggingface.co/datasets/bigcode/starcoderdata) (290 GB) |
| Combined Dataset Size | Around 950B tokens |
| Total Tokens During Training | 3 trillion |
| Name | Description |
|-------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------------------------|
| Parameters | 1.1B |
| Model Size | Layers: 22, Heads: 32, Query Groups: 4, Embedding Size: 2048, Intermediate Size: 5632 |
| Sequence Length | 2048 |
| Learning Rate | 4e-4 |
| Learning Rate Schedule | Cosine with 2000 warmup steps |
| Training Data | [SlimPajama](https://huggingface.co/datasets/cerebras/slimpajama-627b) (893 GB), [Starcoder](https://huggingface.co/datasets/bigcode/starcoderdata) (290 GB) |
| Combined Dataset Size | Around 950B tokens |
| Total Tokens During Training | 3 trillion (3 epochs) |
| Time to complete training | ~ 4 weeks with 64 A100 GPUs |
| Model FLOPs Utilization (MFU) | 52% |

(this table was sourced from the author's [README](https://github.com/jzhang38/TinyLlama/))

Expand Down