README.md

---
license: mit
datasets:
- pints-ai/Expository-Prose-V1
- HuggingFaceH4/ultrachat_200k
- Open-Orca/SlimOrca-Dedup
- meta-math/MetaMathQA
- HuggingFaceH4/deita-10k-v0-sft
- WizardLM/WizardLM_evol_instruct_V2_196k
- togethercomputer/llama-instruct
- LDJnr/Capybara
- HuggingFaceH4/ultrafeedback_binarized
language:
- en
model-index:
- name: 1.5-Pints
  results:
  - task:
      type: text-generation
    dataset:
      name: MTBench
      type: ai2_arc
    metrics:
    - name: MTBench
      type: LLM-as-a-Judge
      value: 3.4
    source:
      name: MTBench
      url: https://huggingface.co/spaces/lmsys/mt-bench
pipeline_tag: text-generation
---

# 1.5-Pints -- A model pretrained in 9 days by using high quality data

## How to use

**Build LlamaCPP**
Refer to https://github.com/ggerganov/llama.cpp/blob/master/docs/build.md on how to build.

**Download Model**
```bash
git clone https://huggingface.co/pints-ai/1.5-Pints-16K-v0.1-GGUF --local-dir PATH/TO/MODEL
```

**Usage**

```bash
# FP32
./llama-cli --model PATH/TO/MODEL/1.5-Pints-16K-v0.1-fp32.gguf --n-gpu-layers 999 --repeat-penalty 1.3 --prompt "Predict what life will be like 100 years from now." 

# FP16
./llama-cli --model PATH/TO/MODEL/1.5-Pints-16K-v0.1-fp16.gguf --n-gpu-layers 999 --repeat-penalty 1.3 --prompt "Predict what life will be like 100 years from now." 
```
`Note`: As at time of publish, `bf16` is slow on llama.cpp (CUDA), hence not recommended for use.

**Compute Infrastructure**<br>
This model can be served with a GPU containing at least 8GB of VRAM.
<br><br>

## Description
1.5 Pints is a Large Language Model that significantly advances the efficiency of LLM training by emphasizing data quality over quantity. Our [pre-training corpus](https://huggingface.co/datasets/pints-ai/Expository-Prose-V1) is a meticulously curated dataset of 57 billion tokens, thus making pre-training more accessible and environmentally-friendly.
<br><br>

## Results
**MTBench**<br>
[MTBench](https://huggingface.co/spaces/lmsys/mt-bench) is a popular evaluation harness that uses strong LLMs like GPT-4 to act as judges and assess the quality of the models' responses./
| Model | Score | Parameter Size | Pretrain Tokens |
|:-:|:-:|:-:|:-:|
| meta-llama/Llama-2-7b-chat-hf | 6.27 | 7B | 2T |
| microsoft/phi-2 | 5.83 | 2.7B | 1.4T |
| google/gemma-2b-it | 5.44 | 2B | 3T |
| stabilityai/stablelm-2-1_6b-chat | 4.7 | 1.6B | 2T |
| **1.5-Pints-2K** | **3.73** | **1.57B** | **0.115T** |
| TinyLlama/TinyLlama-1.1B-Chat-v1.0 | 3.72 | 1.1B | 3T |
| **1.5-Pints-16K** | **3.40** | **1.57B** | **0.115T** |
| apple/OpenELM-1_1B-Instruct | 3.34 | 1B | 1.8T |
| microsoft/phi-1_5 | 3.33 | 1.3B | 0.15T |
| databricks/dolly-v2-3b | 2.33 | 3B | 0.3T |
| EleutherAI/pythia-2.8b | 1.81 | 2.8B | 0.3T |
| tiiuae/falcon-rw-1b | 1.18 | 1B | 0.35T |
<br><br>

The 2K context window version of 1.5-Pints can be found [here](https://huggingface.co/pints-ai/1.5-Pints-2K-v0.1-GGUF).

## Technical Specifications
**Architecture**<br>
Llama 2 Autoregressive Model with **16K Context Window** and Mistral tokenizer. The model uses Float32 precision.

| Parameters | Vocab Size | Embedding Size | Context Length | Layers | Heads | Query Groups | Intermediate Hidden Size |
|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|
| 1,565,886,464 | 32,064 | 2,048 | 16,384 | 24 | 32 | 4 | 8,192 |

**Context Lengths**<br>
1.5-Pints comes in 2 context lengths - 16k (16,384) and 2k (2,048).

**Prompt template**<br>
This model has been finetuned and preference-optimized using the ChatML template.
```
<|im_start|>system 
{SYSTEM_PROMPT}<|im_end|> 
<|im_start|>user 
{PROMPT}<|im_end|> 
<|im_start|>assistant 
```
<br><br>

## Uses
**Direct Use**<br>
This model is meant to be an efficient and fine-tunable helpful assistant. It is designed to excel in user assistance and reasoning, and rely less on internal knowledge and factuals. Thus, for knowledge retrieval purposes, it should be used with Retrieval Augmented Generation.

**Downstream Use**<br>
Given the size of this model, it is possible to launch multiple instances of it for use in agentic context without breaking the compute bank.

**Recommendations**<br>
- It is recommended to finetune this model for domain adaption, and use it for a specialized tasks.
- To reap full performance, use a repetition penalty of 1.3 rather than 1.
<br><br>

## Training Data
**Pre-Train Data**<br>
Dataset: [pints-ai/Expository-Prose-V1](https://huggingface.co/datasets/pints-ai/Expository-Prose-V1)


**Fine-Tune Data**<br>
Corpora:
- [HuggingFaceH4/ultrachat](https://huggingface.co/datasets/HuggingFaceH4/ultrachat_200k)
- [Open-Orca/SlimOrca-Dedup](https://huggingface.co/datasets/Open-Orca/SlimOrca-Dedup)
- [meta-math/MetaMathQA](https://huggingface.co/datasets/meta-math/MetaMathQA)
- [HuggingFaceH4/deita-10k-v0-sft](https://huggingface.co/datasets/HuggingFaceH4/deita-10k-v0-sft)
- [WizardLM/WizardLM_evol_instruct_V2_196k](https://huggingface.co/datasets/WizardLMTeam/WizardLM_evol_instruct_V2_196k)
- [togethercomputer/llama-instruct](https://huggingface.co/togethercomputer/Llama-2-7B-32K-Instruct)
- [LDJnr/Capybara](https://huggingface.co/datasets/LDJnr/Capybara)

**DPO Data**<br>
Dataset: [HuggingFaceH4/ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized)
<br><br>

## Training Procedure
Both Pre-Train and Finetuning used [our fork](https://github.com/Pints-AI/1.5-Pints) of the [LitGPT Framework](https://github.com/Lightning-AI/litgpt). For DPO, we used the methods set out in [The Alignment Handbook](https://github.com/huggingface/alignment-handbook/blob/main/scripts/run_dpo.py). More details can be found in our [paper](TOBEADDED).

## Training Hyperparameters
**Pre-Train**<br>
| Hyperparameter | Value |
|:-:|:-:|
| Optimizer | AdamW(Beta1=0.9, Beta2=0.95) |
| Learning Rate Scheduler | Cosine |
| Max Learning Rate | 4.0x10-4 |
| Min Learning Rate | 4.0x10-5 |
| Warmup Steps | 2,000 |
| Batch Size | 2,097,152 |
| Weight Decay | 0.1 |
| Gradient Clipping Threshold | 1.0 |

**SFT**<br>
| Hyperparameter | Value |
|:-:|:-:|
| Optimizer | AdamW(Beta1=0.9, Beta2=0.95) |
| Warmup steps | 1,126 (10%)
| Peak learning rate | 2e-5 |
| Learning rate scheduler | Cosine |
| Weight Decay | 0.1 |

**DPO**<br>
DPO parameters used are the exact same as those specified in [The Alignment Handbook](https://github.com/huggingface/alignment-handbook).
<br><br>

## Citation
**Attribution**
- **Developed by:** [calvintwr](https://huggingface.co/calvintwr), [lemousehunter](https://huggingface.co/lemousehunter)
- **Funded by** [PintsAI](https://pints.ai/)
- **Released by:** [PintsAI](https://pints.ai/)
- **Model type:** Large Language Model
- **Language(s) (NLP):** English
- **License:** [MIT License](https://opensource.org/license/mit)
<br><br>

**BibTeX:**
[More Information Needed]

**APA**
[More Information Needed]
<br><br>

## Legal Warning
Though best efforts has been made to ensure, as much as possible, that all texts in the training corpora are royalty free, this does not constitute a legal guarantee that such is the case. **By using any of the models, corpora or part thereof, the user agrees to bear full responsibility to do the necessary due diligence to ensure that he / she is in compliance with their local copyright laws**. 

Additionally, the **user agrees to bear any damages** arising as a direct cause (or otherwise) of using any artifacts released by the pints research team, as well as full responsibility for the consequences of his / her usage (or implementation) of any such released artifacts. The user also indemnifies Pints Research Team (and any of its members or agents) of any damage, related or unrelated, to the release or subsequent usage of any findings, artifacts or code by the team.

For the avoidance of doubt, **any artifacts released by the Pints Research team are done so in accordance with the "fair use"** clause of Copyright Law, in hopes that this will aid the research community in bringing LLMs to the next frontier.