* feat(llama.cpp): Enable decentralized, distributed inference
As https://github.com/mudler/LocalAI/pull/2324 introduced distributed inferencing thanks to
@rgerganov implementation in https://github.com/ggerganov/llama.cpp/pull/6829 in upstream llama.cpp, now
it is possible to distribute the workload to remote llama.cpp gRPC server.
This changeset now uses mudler/edgevpn to establish a secure, distributed network between the nodes using a shared token.
The token is generated automatically when starting the server with the `--p2p` flag, and can be used by starting the workers
with `local-ai worker p2p-llama-cpp-rpc` by passing the token via environment variable (TOKEN) or with args (--token).
As per how mudler/edgevpn works, a network is established between the server and the workers with dht and mdns discovery protocols,
the llama.cpp rpc server is automatically started and exposed to the underlying p2p network so the API server can connect on.
When the HTTP server is started, it will discover the workers in the network and automatically create the port-forwards to the service locally.
Then llama.cpp is configured to use the services.
This feature is behind the "p2p" GO_FLAGS
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* go mod tidy
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* ci: add p2p tag
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* better message
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat(functions): allow to use JSONRegexMatch unconditionally
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat(functions): make json_regex_match a list
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Correct llama3-8b-instruct model file
This must be a mistake because the config tries to use a model file that is different from the one actually being downloaded.
I assumed the downloaded file is what should be used so I corrected the specified model file to that
Signed-off-by: Aleksandr Oleinikov <10602045+tannisroot@users.noreply.github.com>
feat(functions): support mixed JSON BNF grammar
This PR provides new options to control how functions are extracted from
the LLM, and also provides more control on how JSON grammars can be used
(also in conjunction).
New YAML settings introduced:
- `grammar_message`: when enabled, the generated grammar can also decide
to push strings and not only JSON objects. This allows the LLM to pick
to either respond freely or using JSON.
- `grammar_prefix`: Allows to prefix a string to the JSON grammar
definition.
- `replace_results`: Is a map that allows to replace strings in the LLM
result.
As an example, consider the following settings for Hermes-2-Pro-Mistral,
which allow extracting both JSON results coming from the model, and the
ones coming from the grammar:
```yaml
function:
# disable injecting the "answer" tool
disable_no_action: true
# This allows the grammar to also return messages
grammar_message: true
# Suffix to add to the grammar
grammar_prefix: '<tool_call>\n'
return_name_in_function_response: true
# Without grammar uncomment the lines below
# Warning: this is relying only on the capability of the
# LLM model to generate the correct function call.
# no_grammar: true
# json_regex_match: "(?s)<tool_call>(.*?)</tool_call>"
replace_results:
"<tool_call>": ""
"\'": "\""
```
Note: To disable entirely grammars usage in the example above, uncomment the
`no_grammar` and `json_regex_match`.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* auto select cpu variant
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* remove cuda target for now
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* fix metal
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* fix path
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* cuda
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* auto select cuda
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* update test
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* select CUDA backend only if present
Signed-off-by: mudler <mudler@localai.io>
* ci: keep cuda bin in path
Signed-off-by: mudler <mudler@localai.io>
* Makefile: make dist now builds also cuda
Signed-off-by: mudler <mudler@localai.io>
* Keep pushing fallback in case auto-flagset/nvidia fails
There could be other reasons for which the default binary may fail. For example we might have detected an Nvidia GPU,
however the user might not have the drivers/cuda libraries installed in the system, and so it would fail to start.
We keep the fallback of llama.cpp at the end of the llama.cpp backends to try to fallback loading in case things go wrong
Signed-off-by: mudler <mudler@localai.io>
* Do not build cuda on MacOS
Signed-off-by: mudler <mudler@localai.io>
* cleanup
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
* Apply suggestions from code review
Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
---------
Signed-off-by: Sertac Ozercan <sozercan@gmail.com>
Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
Signed-off-by: mudler <mudler@localai.io>
Co-authored-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
Co-authored-by: mudler <mudler@localai.io>
When enabling grammar with functions, it might be useful to
allow more flexibility to support models that are fine-tuned against returning
function calls of the form of { "name": "function_name", "arguments" {...} }
rather then { "function": "function_name", "arguments": {..} }.
This might call out to a more generic approach later on, but for the moment being we can easily support both
as we have just to specific different types.
If needed we can expand on this later on
Signed-off-by: mudler <mudler@localai.io>
