| --- |
| license: apache-2.0 |
| language: |
| - en |
| - ko |
| - ja |
| --- |
| # Intermediate Checkpoints Release |
|
|
| For the first time among Korean-targeted LLMs, we’re releasing **intermediate checkpoints** from the Tri family—**0.5B**, **1.9B**, and **7B**—to advance research on LLM training dynamics. We release checkpoints at regular step intervals— **≈20B tokens (0.5B), ≈40B (1.9B), and ≈160B (7B & 70B)** —enabling consistent analysis of training dynamics. Each step’s release is distinguished by its **branch name**. |
| We’re also sharing the **0.5B** and **1.9B** runs—originally produced for system bring-up but now available as valuable artifacts for analyzing training behavior at smaller scales. |
|
|
| You can browse all intermediate checkpoints here: |
| - **Tri-0.5B** → [https://huggingface.co/trillionlabs/0.5B-Intermediate-Checkpoints](https://huggingface.co/trillionlabs/0.5B-Intermediate-Checkpoints) |
| - **Tri-1.9B** → [https://huggingface.co/trillionlabs/1.9B-Intermediate-Checkpoints](https://huggingface.co/trillionlabs/1.9B-Intermediate-Checkpoints) |
| - **Tri-7B** → [https://huggingface.co/trillionlabs/Tri-7B-Intermediate-Checkpoints](https://huggingface.co/trillionlabs/Tri-7B-Intermediate-Checkpoints) |
| - **Tri-70B(SFT Preview)** → [https://huggingface.co/trillionlabs/Tri-70B-Intermediate-Checkpoints](https://huggingface.co/trillionlabs/Tri-70B-Intermediate-Checkpoints) |
| |
| Feel free to check out the full Tri-series collection here: |
| - https://huggingface.co/collections/trillionlabs/tri-series-687fa9ff7eb23e8ba847ef93 |
|
|
| Dive into the full details—including training configuration and loss curves —on our [blog](https://trillionlabs.co/research/tri-series-intermediate-checkpoints-release). |
|
|
|
|
|
|
| # Usage |
|
|
| ```python |
| from transformers import AutoModelForCausalLM, AutoTokenizer |
| |
| INTERMEDIATE_STEP = "0000020000" |
| model = AutoModelForCausalLM.from_pretrained('trillionlabs/0.5B-Intermediate-Checkpoints', revision=INTERMEDIATE_STEP, trust_remote_code=True) |
| tokenizer = AutoTokenizer.from_pretrained('trillionlabs/0.5B-Intermediate-Checkpoints', revision=INTERMEDIATE_STEP, trust_remote_code=True) |
| |
| ... |
| ``` |
|
|