README.md

---
license: bigscience-openrail-m
library_name: transformers
tags:
- code
- gpt_bigcode
datasets:
- nuprl/MultiPL-T
metrics:
- code_eval
model-index:
- name: MultiPLCoder-1b-OCaml
  results:
  - task:
      type: text-generation
    dataset:
      name: MultiPL-HumanEval (Lua)
      type: nuprl/MultiPL-E
    metrics:
    - type: pass@1
      value: 0.173
      name: pass@1
      verified: true
    - type: pass@1
      value: 0.113
      name: pass@1
      verified: true
    - type: pass@1
      value: 0.097
      name: pass@1
      verified: true
---
# MultiPLCoder-1b

1 billion parameter version of MultiPLCoder, a set of StarCoder-based models finetuned on the [MultiPL-T dataset](https://huggingface.co/datasets/nuprl/MultiPL-T).
These models are state-of-the-art at low-resource languages, such as: Lua, Racket, and OCaml.

This 1 billion parameter model is small enough such that it can generate code on the CPU.
For a larger, more capable model, check out [MultiPLCoder-15b](https://huggingface.co/nuprl/MultiPLCoder-15b).


## Language Revision Index

This is the revision index for the best-performing models for their respective langauge.

| Langauge      | Revision ID | Epoch |
| ------------- | ----------- | ----- |
| Lua           | `7e96d931547e342ad0661cdd91236fe4ccf52545`         | 3    |
| Racket        | `2cdc541bee1db4da80c0b43384b0d6a0cacca5b2`         | 5    |
| OCaml         | `e8a24f9e2149cbda8c3cca264a53c2b361b7a031`         | 6    |

## Usage

To utilize one of the models in this repository, you must first select a commit revision for that model from the table above.
For example, to use the Lua model:
```py
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("nuprl/MultiPLCoder-1b")
lua_revision="7e96d931547e342ad0661cdd91236fe4ccf52545"
model = AutoModelForCausalLM.from_pretrained("nuprl/MultiPLCoder-1b", revision=lua_revision)
```

Note that the model's default configuration does not enable caching, therefore you must specify to use the cache on generation.
```py
toks = tokenizer.encode("-- Hello World", return_tensors="pt")
out = model.generate(toks, use_cache=True,  do_sample=True, temperature=0.2, top_p=0.95, max_length=50)
print(tokenizer.decode(out[0], skip_special_tokens=True))
```
```
-- Hello World!
-- :param name: The name of the person to say hello to
-- :return: A greeting
local function say_hello(name)
  return "Hello ".. name
end
```