Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Docs 1.5 (prompt_fn) #595

Merged
merged 3 commits into from
Sep 22, 2023
Merged

Docs 1.5 (prompt_fn) #595

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
48a82af
Update settings.md
aaazzam Sep 22, 2023
f011b84
prompt-fn docs
aaazzam Sep 22, 2023
18bbbe4
change docs
aaazzam Sep 22, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 1 addition & 3 deletions docs/components/ai_classifier.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,10 +194,8 @@ By default, `@ai_classifier` uses the global LLM settings. To specify a particul


```python
from marvin.engine.language_models import chat_llm


@ai_classifier(model=chat_llm("openai/gpt-3.5-turbo-0613"))
@ai_classifier(model="openai/gpt-3.5-turbo-0613", temperature = 0)
class Sentiment(Enum):
POSITIVE = 1
NEGATIVE = -1
Expand Down
13 changes: 3 additions & 10 deletions docs/components/ai_model.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -143,26 +143,19 @@ Note that the kwarg is `model_` with a trailing underscore; this is to avoid con


```python
from marvin.engine.language_models import chat_llm


@ai_model(model=chat_llm(model="openai/gpt-3.5-turbo", temperature=0))
@ai_model(model="openai/gpt-3.5-turbo", temperature=0)
class Location(BaseModel):
city: str
state: str


print(Location("The Big Apple"))
print(
Location(
"The Big Apple",
model_=chat_llm(model="openai/gpt-3.5-turbo", temperature=1),
)
)

```

city='New York' state='New York'
city='New York' state='New York'



## Features
Expand Down
3 changes: 1 addition & 2 deletions docs/components/overview.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,9 +10,8 @@ Marvin introduces a number of components that can become the building blocks of


```python
from marvin.engine.language_models.openai import OpenAIChatLLM

@ai_fn(model=OpenAIChatLLM(model="openai/gpt-3.5-turbo-16k"))
@ai_fn(model="openai/gpt-3.5-turbo-16k", temperature = 0)
def my_ai_fn():
"""..."""
```
Expand Down
2 changes: 1 addition & 1 deletion docs/configuration/settings.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,7 +39,7 @@ Valid `llm_model` settings are strings with the form `"{provider_key}/{model_nam
| --- | --- | --- | --- |
| OpenAI | `openai` | `gpt-3.5-turbo`, `gpt-4` (default), or any other [compatible model](https://platform.openai.com/docs/models/) | Marvin is generally tested and optimized with this provider. |
| Anthropic | `anthropic` | `claude-2`, `claude-instant-1` or any other [compatible model](https://docs.anthropic.com/claude/reference/selecting-a-model) | Available in Marvin 1.1|
| Azure OpenAI Service | `azure_openai` | `gpt-35-turbo`, `gpt-4`, or any other [compatible model](https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models) | The Azure OpenAI Service shares all the same configuration options as the OpenAI models, as well as a few additional ones. Available in Marvin 1.1. |
| Azure OpenAI Service | `azure_openai` | `gpt-3.5-turbo`, `gpt-4`, or any other [compatible model](https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models) | The Azure OpenAI Service shares all the same configuration options as the OpenAI models, as well as a few additional ones. Available in Marvin 1.1. |

## LLM Configuration

Expand Down
72 changes: 38 additions & 34 deletions docs/llms/llms.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,13 +38,13 @@ specify a model name.
# Call claude-2 simply by specifying it inside of ChatCompletion.
anthropic = ChatCompletion('claude-2').create(messages = messages)
```
We can now access both results as we would with OpenAI.
We can now access both results as we would with OpenAI (after calling .response)

```python
print(openai.choices[0].message.content)
print(openai.response.choices[0].message.content)
# Hello! I'm an AI, so I don't have feelings, but I'm here to help you. How can I assist you?

print(anthropic.choices[0].message.content)
print(anthropic.response.choices[0].message.content)
# I'm doing well, thanks for asking!
```

Expand Down Expand Up @@ -87,7 +87,7 @@ keyword arguments passed to ChatCompletion will be persisted and passed to subse
from marvin import openai


openai.ChatCompletion.create(
openai.ChatCompletion().create(
messages = [{
'role': 'user',
'content': 'Hey! How are you?'
Expand Down Expand Up @@ -124,7 +124,7 @@ Let's consider two examples.
with_sugar: bool = False


response = openai.ChatCompletion.create(
response = openai.ChatCompletion().create(
messages = [{
'role': 'user',
'content': 'Can I get a small soymilk latte?'
Expand Down Expand Up @@ -154,7 +154,7 @@ Let's consider two examples.
swedish: str


response = openai.ChatCompletion.create(
response = openai.ChatCompletion().create(
messages = [
{
'role': 'system',
Expand Down Expand Up @@ -214,7 +214,7 @@ Let's consider an example.

```python

response = openai.ChatCompletion.create(
response = ZZZcreate(
messages = [{
'role': 'user',
'content': 'What if I put it $100 every month for 60 months at 12%?'
Expand Down Expand Up @@ -348,11 +348,11 @@ Let's consider an example.
conversation.send(messages = [{'role': 'user', 'content': prompt}])

# While the most recent turn has a function call, evaluate it.
while conversation.last_response.has_function_call():
while conversation.last_turn.has_function_call():

# Send the most recent function call to the conversation.
conversation.send(messages = [
conversation.last_response.call_function()
conversation.last_turn.call_function()
])

```
Expand All @@ -361,7 +361,7 @@ Let's consider an example.

```python

conversation.last_response.choices[0].message.content
conversation.last_turn.choices[0].message.content

# The result of adding 4124124 and 424242 is 4548366. When this result is divided by 48124,
# the answer is approximately 94.51346521486161.
Expand All @@ -371,29 +371,33 @@ Let's consider an example.
If we want to see the entire state, every `[request, response]` pair is held in the conversation's
`turns`.
```python
[response.choices[0].message for response in conversation.turns]

# [<OpenAIObject at 0x120667c50> JSON: {
# "role": "assistant",
# "content": null,
# "function_call": {
# "name": "add",
# "arguments": "{\n \"x\": 4124124,\n \"y\": 424242\n}"
# }
# },
# <OpenAIObject at 0x1206f4830> JSON: {
# "role": "assistant",
# "content": null,
# "function_call": {
# "name": "divide",
# "arguments": "{\n \"x\": 4548366,\n \"y\": 48124\n}"
# }
# },
# <OpenAIObject at 0x1206f4b90> JSON: {
# "role": "assistant",
# "content": "The result of adding 4124124 and 424242 is 4548366.
# When this result is divided by 48124, the answer is
# approximately 94.51346521486161."
# }]
[turn.response.choices[0].message.dict() for turn in conversation.turns]

[
{
"content": null,
"role": "assistant",
"name": null,
"function_call": {
"name": "add",
"arguments": "{\n \"x\": 4124124,\n \"y\": 424242\n}"
}
},
{
"content": null,
"role": "assistant",
"name": null,
"function_call": {
"name": "divide",
"arguments": "{\n \"x\": 4548366,\n \"y\": 48124\n}"
}
},
{
"content": "4124124 + 424242 divided by 48124 is approximately 94.51346521486161.",
"role": "assistant",
"name": null,
"function_call": null
}
]

```
Loading