vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

server : accept extra_context for the infill endpoint (#9874)

Browse source 
* server : accept extra_context for the infill endpoint

ggml-ci

* server : update readme [no ci]

* server : use repo-level FIM pattern if possible

ggml-ci
This commit is contained in:
Georgi Gerganov 2024-10-13 21:31:35 +03:00 • committed by GitHub
parent c7181bd294
commit d4c19c0f5c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 153 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

21
examples/server/README.md
View file

@ -524,9 +524,30 @@ Takes a prefix and a suffix and returns the predicted completion as stream.
- `input_prefix`: Set the prefix of the code to infill.
- `input_suffix`: Set the suffix of the code to infill.
- `prompt`: Added after the `FIM_MID` token
- `extra_context`: Additional context inserted before the FIM prefix. See https://github.com/ggerganov/llama.cpp/pull/9874
It also accepts all the options of `/completion`.
If the model has `FIM_REPO` and `FIM_FILE_SEP` tokens, the [repo-level pattern](https://arxiv.org/pdf/2409.12186) is used:
```txt
<FIM_REP>myproject
<FIM_SEP>{chunk 0 filename}
{chunk 0 text}
<FIM_SEP>{chunk 1 filename}
{chunk 1 text}
...
<FIM_SEP>filename
<FIM_PRE>[input_prefix]<FIM_SUF>[input_suffix]<FIM_MID>[prompt]
```
If the tokens are missing, then the extra context is simply prefixed at the start:
```txt
[extra_context]<FIM_PRE>[input_prefix]<FIM_SUF>[input_suffix]<FIM_MID>[prompt]
```
### **GET** `/props`: Get server global properties.
This endpoint is public (no API key check). By default, it is read-only. To make POST request to change global properties, you need to start server with `--props`