---
library_name: transformers
license: other
base_model: dogtooth/open-lm-3b-202301
tags:
- llama-factory
- full
- sft
- chrononauts
- temporal-lm
model-index:
- name: open-lm-3b-202301-sft-nontemporal
  results: []
---

# Open LM 3B January 2023 — SFT (Non-Temporal)

This model is a supervised fine-tuned (SFT) version of [dogtooth/open-lm-3b-202301](https://huggingface.co/dogtooth/open-lm-3b-202301), trained on the non-temporal split of the [mattwang123/chrononauts-sft-filtered](https://huggingface.co/datasets/mattwang123/chrononauts-sft-filtered) dataset.

It is part of the **Chrononauts** project (JHU CLSP), which studies temporal/time-continual language modeling using Apple Open LM models with different knowledge cutoff dates. Based on the [TiC-LM paper](https://arxiv.org/abs/2410.14660).

## Model Details

| Property | Value |
|----------|-------|
| **Base model** | [dogtooth/open-lm-3b-202301](https://huggingface.co/dogtooth/open-lm-3b-202301) |
| **Parameters** | 2.7B |
| **Architecture** | LLaMA-style (pre-norm, SwiGLU, RoPE) |
| **Knowledge cutoff** | January 2023 |
| **Context length** | 2,048 tokens |
| **Vocab size** | 50,432 |
| **Training data** | [mattwang123/chrononauts-sft-filtered](https://huggingface.co/datasets/mattwang123/chrononauts-sft-filtered) (non_temporal split) |
| **Eval data** | HellaSwag |
| **Train loss** | 0.923 |
| **Eval loss** | 2.602 |

## Chat Template

This model uses a simple `Human:` / `Assistant:` prompt format with `<|endoftext|>` as the separator token.

### Format

```
Human: {user_message}<|endoftext|>
Assistant:{assistant_response}<|endoftext|>
```

For multi-turn conversations:

```
Human: {first_message}<|endoftext|>
Assistant:{first_response}<|endoftext|>
Human: {second_message}<|endoftext|>
Assistant:{second_response}<|endoftext|>
```

With a system prompt:

```
System: {system_message}<|endoftext|>
Human: {user_message}<|endoftext|>
Assistant:{assistant_response}<|endoftext|>
```

### Usage with Transformers

```python
from transformers import AutoTokenizer, AutoModelForCausalLM

model_id = "dogtooth/open-lm-3b-202301-sft-nontemporal"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)

# Using the chat template
messages = [
    {"role": "user", "content": "What is the capital of France?"}
]
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
# Result: "Human: What is the capital of France?<|endoftext|>\nAssistant:"

inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=False))
```

### Manual Prompting

```python
prompt = "Human: What is the capital of France?<|endoftext|>\nAssistant:"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=False))
```

## Training Hyperparameters

- **Learning rate:** 1e-5
- **Batch size:** 2 per device x 4 GPUs x 4 gradient accumulation = 32 effective
- **Epochs:** 3
- **Scheduler:** Cosine with 10% warmup
- **Precision:** bf16
- **DeepSpeed:** ZeRO Stage 2
- **Hardware:** 4x NVIDIA A100 80GB

## Framework Versions

- Transformers 4.57.1
- PyTorch 2.6.0+cu124
- Datasets 4.0.0
- Tokenizers 0.22.2