# Guide to Hugging Face CodeLlama 70B Instruct Model

Hugging Face‘s Code Llama is a collection of pretrained and fine-tuned generative text models, and this guide is specifically for the 70B instruct-tuned version. This model is designed for general code synthesis and understanding.

## Model Capabilities

### Model Use

To use the 70B Instruct model, you can start by installing the `transformers` library using the following command:

“`bash
pip install transformers accelerate
“`

#### Chat Use:

To use the model for chat, you can use the built-in chat template provided by the `transformers` library. Here’s an example of how to use it:

“`python
from transformers import AutoTokenizer, AutoModelForCausalLM
import transformers
import torch

model_id = “codellama/CodeLlama-70b-Instruct-hf”
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.float16,
device_map=”auto”,
)

chat = [
{“role”: “system”, “content”: “You are a helpful and honest code assistant expert in JavaScript. Please, provide all answers to programming questions in JavaScript”},
{“role”: “user”, “content”: “Write a function that computes the set of sums of all contiguous sublists of a given list.”},
]
inputs = tokenizer.apply_chat_template(chat, return_tensors=”pt”).to(“cuda”)

output = model.generate(input_ids=inputs, max_new_tokens=200)
output = output[0].to(“cpu”)
print(tokenizer.decode(output))
“`

You can also use the model for text or code completion using transformers’ pipeline interface.

#### Chat Prompt

The CodeLlama 70B Instruct model uses a different format for the chat prompt than previous Llama 2 or CodeLlama models. You can use the tokenizer’s chat template to build the chat prompt easily.

## Model Details

* Note: Use of this model is governed by the Meta license. Meta developed and publicly released the Code Llama family of large language models (LLMs).

### Model Developers
Meta

### Variations
Code Llama comes in multiple model sizes and variants:
– Code Llama
– Code Llama – Python
– Code Llama – Instruct

This repository specifically contains the Instruct version of the 70B parameters model.

### Input / Output
The model’s input and output are text only.

### Model Architecture
Code Llama is an auto-regressive language model that uses an optimized transformer architecture. It was fine-tuned with up to 16k tokens and does not support long context of up to 100k tokens.

### Model Dates
Code Llama and its variants have been trained between January 2023 and January 2024.

### Status
This is a static model trained on an offline dataset. Future versions of Code Llama – Instruct will be released as Meta improves model safety with community feedback.

### License
A custom commercial license is available at: [Meta AI Resources](https://ai.meta.com/resources/models-and-libraries/llama-downloads/)

### Research Paper
More information can be found in the paper “Code Llama: Open Foundation Models for Code” or on its [arXiv page](https://arxiv.org/abs/2308.12950).

## Intended Use

### Intended Use Cases
Code Llama and its variants are intended for commercial and research use in English and relevant programming languages.

### Out-of-Scope Uses
Use in any manner that violates applicable laws or regulations, use in languages other than English, and any other use that is prohibited by the Acceptable Use Policy and Licensing Agreement for Code Llama and its variants are considered out-of-scope.

## Hardware and Software

### Training Factors
The training and fine-tuning of the released models have been performed on Meta’s Research Super Cluster.

### Carbon Footprint
In aggregate, training all 12 Code Llama models required 1400K GPU hours of computation on hardware of type A100-80GB (TDP of 350-400W). Estimated total emissions were 228.55 tCO2eq, 100% of which were offset by Meta’s sustainability program.

## Evaluation Results

See evaluations for the main models and detailed ablations in Section 3 and safety evaluations in Section 4 of the research paper.

## Ethical Considerations and Limitations

Code Llama and its variants are a new technology that carries risks with use. The potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate or objectionable responses. Therefore, before deploying any applications of Code Llama, developers should perform safety testing and tuning tailored to their specific applications of the model.

Please refer to the Responsible Use Guide available at [Meta AI Responsible Use Guide](https://ai.meta.com/llama/responsible-use-guide) for more information.

For more details and links to other models, you can visit the original repository.

Source link
# Hugging Face Tutorial for Code Llama Model

Hugging Face is a platform that provides pretrained and fine-tuned generative text models. In this tutorial, we will cover the usage of the Code Llama model, specifically the 70B instruct-tuned version in the Hugging Face Transformers format.

## Model Installation

To use the Code Llama 70B Instruct model with Hugging Face, you need to install the `transformers` package. You can do this using pip:

“`bash
pip install transformers accelerate
“`

## Model Use

The 70B Instruct model can be used for chat interactions or for text/code generation using the Python programming language.

### Chat Use

To use the model for chat interactions, you can use the built-in chat template provided by the `transformers` library:

“`python
from transformers import AutoTokenizer, AutoModelForCausalLM
import transformers
import torch

model_id = “codellama/CodeLlama-70b-Instruct-hf”
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.float16,
device_map=”auto”,
)

chat = [
{“role”: “system”, “content”: “You are a helpful and honest code assistant expert in JavaScript. Please, provide all answers to programming questions in JavaScript”},
{“role”: “user”, “content”: “Write a function that computes the set of sums of all contiguous sublists of a given list.”},
]
inputs = tokenizer.apply_chat_template(chat, return_tensors=”pt”).to(“cuda”)

output = model.generate(input_ids=inputs, max_new_tokens=200)
output = output[0].to(“cpu”)
print(tokenizer.decode(output))
“`

### Text/Code Completion

The model can also be used for text or code completion through the `pipeline` interface:

“`python
from transformers import AutoTokenizer
import transformers
import torch

model_id = “codellama/CodeLlama-70b-hf”
tokenizer = AutoTokenizer.from_pretrained(model_id)
pipeline = transformers.pipeline(
“text-generation”,
model=model_id,
torch_dtype=torch.float16,
device_map=”auto”,
)

sequences = pipeline(
‘def fibonacci(‘,
do_sample=True,
temperature=0.2,
top_p=0.9,
num_return_sequences=1,
eos_token_id=tokenizer.eos_token_id,
max_length=100,
)
for seq in sequences:
print(f”Result: {seq[‘generated_text’]}”)
“`

## Chat Prompt

The Chat prompt for CodeLlama 70B Instruct uses a specific format. It includes a system prompt, user queries, and model responses. The chat template provided by the tokenizer simplifies this process. However, you can manually build the chat prompts using Python.

### Manual Chat Prompt

To manually build the chat prompt, you can follow this example:

“`python
chat = [
{“role”: “system”, “content”: “System prompt “},
{“role”: “user”, “content”: “First user query”},
{“role”: “assistant”, “content”: “Model response to first query”},
{“role”: “user”, “content”: “Second user query”},
]
output = “”
for m in chat:
output += f”Source: {m[‘role’]}\n\n {m[‘content’].strip()}”
output += ”
output += “Source: assistant\nDestination: user\n\n ”
output
“`

## Model Details

The Code Llama model comes in different variations and sizes. You are currently using the Instruct version of the 70B parameters model. The model is designed for general code synthesis and understanding tasks.

For more information about the model’s architecture, training dates, license, and research paper, refer to the Hugging Face repository.

## Intended Use

The Code Llama model, including the Instruct variation, is intended for commercial and research use in English and relevant programming languages. However, there are out-of-scope uses that need to be considered.

Please review the intended use cases and out-of-scope uses mentioned in the Hugging Face repository for the Code Llama model.

## Hardware and Software

The model was trained using custom training libraries and Meta’s Research Super Cluster. Details about the training factors and carbon footprint are provided in the Hugging Face repository.

## Evaluation Results

Evaluations for the main models and detailed ablations can be found in Section 3 and safety evaluations in Section 4 of the research paper. Refer to the Hugging Face repository for more information.

## Ethical Considerations and Limitations

The Code Llama model, including its variants, presents ethical considerations and limitations due to the risks associated with its use. It’s important to perform safety testing and tuning tailored to specific applications of the model.

Refer to the Responsible Use Guide available at the provided URL for more information on ethical considerations and limitations.

This tutorial provides a comprehensive guide on using the Code Llama model, specifically the Instruct version, with Hugging Face. For more detailed information, refer to the original Hugging Face repository for the Code Llama model.

Code Llama is a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. This is the repository for the 70B instruct-tuned version in the Hugging Face Transformers format. This model is designed for general code synthesis and understanding. Links to other models can be found in the index at the bottom.

Model capabilities:



Model Use

Install transformers

pip install transformers accelerate

Chat use: The 70B Instruct model uses a different prompt template than the smaller versions. To use it with transformers, we recommend you use the built-in chat template:

from transformers import AutoTokenizer, AutoModelForCausalLM
import transformers
import torch

model_id = "codellama/CodeLlama-70b-Instruct-hf"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
   model_id,
   torch_dtype=torch.float16,
   device_map="auto",
)

chat = [
   {"role": "system", "content": "You are a helpful and honest code assistant expert in JavaScript. Please, provide all answers to programming questions in JavaScript"},
   {"role": "user", "content": "Write a function that computes the set of sums of all contiguous sublists of a given list."},
]
inputs = tokenizer.apply_chat_template(chat, return_tensors="pt").to("cuda")

output = model.generate(input_ids=inputs, max_new_tokens=200)
output = output[0].to("cpu")
print(tokenizer.decode(output))

You can also use the model for text or code completion. This examples uses transformers’ pipeline interface:

from transformers import AutoTokenizer
import transformers
import torch

model_id = "codellama/CodeLlama-70b-hf"
tokenizer = AutoTokenizer.from_pretrained(model_id)
pipeline = transformers.pipeline(
   "text-generation",
   model=model_id,
   torch_dtype=torch.float16,
   device_map="auto",
)

sequences = pipeline(
   'def fibonacci(',
   do_sample=True,
   temperature=0.2,
   top_p=0.9,
   num_return_sequences=1,
   eos_token_id=tokenizer.eos_token_id,
   max_length=100,
)
for seq in sequences:
   print(f"Result: {seq['generated_text']}")



Chat prompt

CodeLlama 70B Instruct uses a different format for the chat prompt than previous Llama 2 or CodeLlama models. As mentioned above, the easiest way to use it is with the help of the tokenizer’s chat template. If you need to build the string or tokens, manually, here’s how to do it.

We’ll do our tests with the following made-up dialog:

chat = [
    {"role": "system", "content": "System prompt    "},
    {"role": "user", "content": "First user query"},
    {"role": "assistant", "content": "Model response to first query"},
    {"role": "user", "content": "Second user query"},
]

First, let’s see what the prompt looks like if we use the chat template:

tokenizer.apply_chat_template(chat, tokenize=False)
'<s>Source: system\n\n System prompt <step> Source: user\n\n First user query <step> Source: assistant\n\n Model response to first query <step> Source: user\n\n Second user query <step> Source: assistant\nDestination: user\n\n '

So each turn of the conversation has a Source (system, user, or assistant), and then the content appears after two newlines and a space. Turns are separated with the special token <step>. After the last turn (which must necessarily come from the user), we invite the model to respond by using the special syntax Source: assistant\nDestination: user\n\n . Let’s see how we can build the same string ourselves:

output = "<s>"
for m in chat:
    output += f"Source: {m['role']}\n\n {m['content'].strip()}"
    output += " <step> "
output += "Source: assistant\nDestination: user\n\n "
output
'<s>Source: system\n\n System prompt <step> Source: user\n\n First user query <step> Source: assistant\n\n Model response to first query <step> Source: user\n\n Second user query <step> Source: assistant\nDestination: user\n\n '

To verify that we got it right, we’ll compare against the reference code in the original GitHub repo. We used the same dialog and tokenized it with the dialog_prompt_tokens function and got the following tokens:

reference_tokens = [1, 7562, 29901, 1788, 13, 13, 2184, 9508, 32015, 7562, 29901, 1404, 13, 13, 3824, 1404, 2346, 32015, 7562, 29901, 20255, 13, 13, 8125, 2933, 304, 937, 2346, 32015, 7562, 29901, 1404, 13, 13, 6440, 1404, 2346, 32015, 7562, 29901, 20255, 13, 14994, 3381, 29901, 1404, 13, 13, 29871]

Let’s see what we get with the string we built using our Python loop. Note that we don’t add “special tokens” because the string already starts with <s>, the beginning of sentence token:

tokens = tokenizer.encode(output, add_special_tokens=False)
assert reference_tokens == tokens

Similarly, let’s verify that the chat template produces the same token sequence:

assert reference_tokens == tokenizer.apply_chat_template(chat)

As a final detail, please note that if the dialog does not start with a system turn, the original code will insert one with an empty content string.



Model Details

*Note: Use of this model is governed by the Meta license. Meta developed and publicly released the Code Llama family of large language models (LLMs).

Model Developers Meta

Variations Code Llama comes in four model sizes, and three variants:

  • Code Llama: base models designed for general code synthesis and understanding
  • Code Llama – Python: designed specifically for Python
  • Code Llama – Instruct: for instruction following and safer deployment

All variants are available in sizes of 7B, 13B, 34B, and 70B parameters.

This repository contains the Instruct version of the 70B parameters model.

Input Models input text only.

Output Models generate text only.

Model Architecture Code Llama is an auto-regressive language model that uses an optimized transformer architecture. It was fine-tuned with up to 16k tokens. This variant does not support long context of up to 100k tokens.

Model Dates Code Llama and its variants have been trained between January 2023 and January 2024.

Status This is a static model trained on an offline dataset. Future versions of Code Llama – Instruct will be released as we improve model safety with community feedback.

License A custom commercial license is available at: https://ai.meta.com/resources/models-and-libraries/llama-downloads/

Research Paper More information can be found in the paper “Code Llama: Open Foundation Models for Code” or its arXiv page.



Intended Use

Intended Use Cases Code Llama and its variants are intended for commercial and research use in English and relevant programming languages. The base model Code Llama can be adapted for a variety of code synthesis and understanding tasks, Code Llama – Python is designed specifically to handle the Python programming language, and Code Llama – Instruct is intended to be safer to use for code assistant and generation applications.

Out-of-Scope Uses Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English. Use in any other way that is prohibited by the Acceptable Use Policy and Licensing Agreement for Code Llama and its variants.



Hardware and Software

Training Factors We used custom training libraries. The training and fine-tuning of the released models have been performed Meta’s Research Super Cluster.
Carbon Footprint In aggregate, training all 12 Code Llama models required 1400K GPU hours of computation on hardware of type A100-80GB (TDP of 350-400W). Estimated total emissions were 228.55 tCO2eq, 100% of which were offset by Meta’s sustainability program.



Evaluation Results

See evaluations for the main models and detailed ablations in Section 3 and safety evaluations in Section 4 of the research paper.



Ethical Considerations and Limitations

Code Llama and its variants are a new technology that carries risks with use. Testing conducted to date has been in English, and has not covered, nor could it cover all scenarios. For these reasons, as with all LLMs, Code Llama’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate or objectionable responses to user prompts. Therefore, before deploying any applications of Code Llama, developers should perform safety testing and tuning tailored to their specific applications of the model.

Please see the Responsible Use Guide available available at https://ai.meta.com/llama/responsible-use-guide.

The

tag in HTML is a generic container for holding all sorts of other elements. In the given HTML snippet, we can see that it is being used to contain various sections of information and instructions about a specific code model – the Code Llama 70B, a pretrained generative text model.

In this context, the

tag is used to structure the information about the model’s capabilities and use cases. The various sections contained within the

include:
– Model Use
– Chat prompt
– Model Details
– Intended Use
– Hardware and Software
– Evaluation Results
– Ethical Considerations and Limitations

Each of these sections provides different use cases and information about how the Code Llama 70B Instruct model can be used in a variety of scenarios related to AI, frameworks, Python, coding, and more.

For example, under the “Model Use” section, there is a block of code illustrating how the model can be used for “Chat use” and for “text or code completion”. This demonstrates that the model can be used as a chatbot or for auto-completion of code or text based on specific prompts.

Similarly, the “Intended Use” section outlines the intended use cases for the model, such as commercial and research use in English and relevant programming languages. It also specifies the out-of-scope uses, mentioning that usage in languages other than English and any violation of applicable laws or regulations is prohibited.

The “Ethical Considerations and Limitations” section raises important ethical considerations about using this model, emphasizing the need for safety testing and tuning tailored to specific applications before deploying any applications of Code Llama.

The

tag has been effectively utilized in this HTML snippet to organize and structure the information about the model and its use cases, providing a clear and concise representation of the capabilities and intended uses of the Code Llama 70B Instruct model.

2024-01-30T11:09:35+01:00