mirror of
https://github.com/hwchase17/langchain
synced 2024-11-06 03:20:49 +00:00
a23c719c8b
3 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Liang Zhang
|
7306600e2f
|
community[patch]: Support SerDe transform functions in Databricks LLM (#16752)
**Description:** Databricks LLM does not support SerDe the
transform_input_fn and transform_output_fn. After saving and loading,
the LLM will be broken. This PR serialize these functions into a hex
string using pickle, and saving the hex string in the yaml file. Using
pickle to serialize a function can be flaky, but this is a simple
workaround that unblocks many use cases. If more sophisticated SerDe is
needed, we can improve it later.
Test:
Added a simple unit test.
I did manual test on Databricks and it works well.
The saved yaml looks like:
```
llm:
_type: databricks
cluster_driver_port: null
cluster_id: null
databricks_uri: databricks
endpoint_name: databricks-mixtral-8x7b-instruct
extra_params: {}
host: e2-dogfood.staging.cloud.databricks.com
max_tokens: null
model_kwargs: null
n: 1
stop: null
task: null
temperature: 0.0
transform_input_fn: 80049520000000000000008c085f5f6d61696e5f5f948c0f7472616e73666f726d5f696e7075749493942e
transform_output_fn: null
```
@baskaryan
```python
from langchain_community.embeddings import DatabricksEmbeddings
from langchain_community.llms import Databricks
from langchain.chains import RetrievalQA
from langchain.document_loaders import TextLoader
from langchain.text_splitter import CharacterTextSplitter
from langchain.vectorstores import FAISS
import mlflow
embeddings = DatabricksEmbeddings(endpoint="databricks-bge-large-en")
def transform_input(**request):
request["messages"] = [
{
"role": "user",
"content": request["prompt"]
}
]
del request["prompt"]
return request
llm = Databricks(endpoint_name="databricks-mixtral-8x7b-instruct", transform_input_fn=transform_input)
persist_dir = "faiss_databricks_embedding"
# Create the vector db, persist the db to a local fs folder
loader = TextLoader("state_of_the_union.txt")
documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
docs = text_splitter.split_documents(documents)
db = FAISS.from_documents(docs, embeddings)
db.save_local(persist_dir)
def load_retriever(persist_directory):
embeddings = DatabricksEmbeddings(endpoint="databricks-bge-large-en")
vectorstore = FAISS.load_local(persist_directory, embeddings)
return vectorstore.as_retriever()
retriever = load_retriever(persist_dir)
retrievalQA = RetrievalQA.from_llm(llm=llm, retriever=retriever)
with mlflow.start_run() as run:
logged_model = mlflow.langchain.log_model(
retrievalQA,
artifact_path="retrieval_qa",
loader_fn=load_retriever,
persist_dir=persist_dir,
)
# Load the retrievalQA chain
loaded_model = mlflow.pyfunc.load_model(logged_model.model_uri)
print(loaded_model.predict([{"query": "What did the president say about Ketanji Brown Jackson"}]))
```
|
||
|
Liang Zhang
|
6479aab74f
|
community[patch]: Add param "task" to Databricks LLM to work around serialization of transform_output_fn (#14933)
**What is the reproduce code?**
```python
from langchain.chains import LLMChain, load_chain
from langchain.llms import Databricks
from langchain.prompts import PromptTemplate
def transform_output(response):
# Extract the answer from the responses.
return str(response["candidates"][0]["text"])
def transform_input(**request):
full_prompt = f"""{request["prompt"]}
Be Concise.
"""
request["prompt"] = full_prompt
return request
chat_model = Databricks(
endpoint_name="llama2-13B-chat-Brambles",
transform_input_fn=transform_input,
transform_output_fn=transform_output,
verbose=True,
)
print(f"Test chat model: {chat_model('What is Apache Spark')}") # This works
llm_chain = LLMChain(llm=chat_model, prompt=PromptTemplate.from_template("{chat_input}"))
llm_chain("colorful socks") # this works
llm_chain.save("databricks_llm_chain.yaml") # transform_input_fn and transform_output_fn are not serialized into the model yaml file
loaded_chain = load_chain("databricks_llm_chain.yaml") # The Databricks LLM is recreated with transform_input_fn=None, transform_output_fn=None.
loaded_chain("colorful socks") # Thus this errors. The transform_output_fn is needed to produce the correct output
```
Error:
```
File "/local_disk0/.ephemeral_nfs/envs/pythonEnv-6c34afab-3473-421d-877f-1ef18930ef4d/lib/python3.10/site-packages/pydantic/v1/main.py", line 341, in __init__
raise validation_error
pydantic.v1.error_wrappers.ValidationError: 1 validation error for Generation
text
str type expected (type=type_error.str)
request payload: {'query': 'What is a databricks notebook?'}'}
```
**What does the error mean?**
When the LLM generates an answer, represented by a Generation data
object. The Generation data object takes a str field called text, e.g.
Generation(text=”blah”). However, the Databricks LLM tried to put a
non-str to text, e.g. Generation(text={“candidates”:[{“text”: “blah”}]})
Thus, pydantic errors.
**Why the output format becomes incorrect after saving and loading the
Databricks LLM?**
Databrick LLM does not support serializing transform_input_fn and
transform_output_fn, so they are not serialized into the model yaml
file. When the Databricks LLM is loaded, it is recreated with
transform_input_fn=None, transform_output_fn=None. Without
transform_output_fn, the output text is not unwrapped, thus errors.
Missing transform_output_fn causes this error.
Missing transform_input_fn causes the additional prompt “Be Concise.” to
be lost after saving and loading.
<!-- Thank you for contributing to LangChain!
Replace this entire comment with:
- **Description:** a description of the change,
- **Issue:** the issue # it fixes (if applicable),
- **Dependencies:** any dependencies required for this change,
- **Tag maintainer:** for a quicker response, tag the relevant
maintainer (see below),
- **Twitter handle:** we announce bigger features on Twitter. If your PR
gets announced, and you'd like a mention, we'll gladly shout you out!
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` to check this
locally.
See contribution guidelines for more information on how to write/run
tests, lint, etc:
https://python.langchain.com/docs/contributing/
If you're adding a new integration, please include:
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in `docs/extras`
directory.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
|
||
Bagatur
|
ed58eeb9c5
|
community[major], core[patch], langchain[patch], experimental[patch]: Create langchain-community (#14463)
Moved the following modules to new package langchain-community in a backwards compatible fashion: ``` mv langchain/langchain/adapters community/langchain_community mv langchain/langchain/callbacks community/langchain_community/callbacks mv langchain/langchain/chat_loaders community/langchain_community mv langchain/langchain/chat_models community/langchain_community mv langchain/langchain/document_loaders community/langchain_community mv langchain/langchain/docstore community/langchain_community mv langchain/langchain/document_transformers community/langchain_community mv langchain/langchain/embeddings community/langchain_community mv langchain/langchain/graphs community/langchain_community mv langchain/langchain/llms community/langchain_community mv langchain/langchain/memory/chat_message_histories community/langchain_community mv langchain/langchain/retrievers community/langchain_community mv langchain/langchain/storage community/langchain_community mv langchain/langchain/tools community/langchain_community mv langchain/langchain/utilities community/langchain_community mv langchain/langchain/vectorstores community/langchain_community mv langchain/langchain/agents/agent_toolkits community/langchain_community mv langchain/langchain/cache.py community/langchain_community mv langchain/langchain/adapters community/langchain_community mv langchain/langchain/callbacks community/langchain_community/callbacks mv langchain/langchain/chat_loaders community/langchain_community mv langchain/langchain/chat_models community/langchain_community mv langchain/langchain/document_loaders community/langchain_community mv langchain/langchain/docstore community/langchain_community mv langchain/langchain/document_transformers community/langchain_community mv langchain/langchain/embeddings community/langchain_community mv langchain/langchain/graphs community/langchain_community mv langchain/langchain/llms community/langchain_community mv langchain/langchain/memory/chat_message_histories community/langchain_community mv langchain/langchain/retrievers community/langchain_community mv langchain/langchain/storage community/langchain_community mv langchain/langchain/tools community/langchain_community mv langchain/langchain/utilities community/langchain_community mv langchain/langchain/vectorstores community/langchain_community mv langchain/langchain/agents/agent_toolkits community/langchain_community mv langchain/langchain/cache.py community/langchain_community ``` Moved the following to core ``` mv langchain/langchain/utils/json_schema.py core/langchain_core/utils mv langchain/langchain/utils/html.py core/langchain_core/utils mv langchain/langchain/utils/strings.py core/langchain_core/utils cat langchain/langchain/utils/env.py >> core/langchain_core/utils/env.py rm langchain/langchain/utils/env.py ``` See .scripts/community_split/script_integrations.sh for all changes |