fix(config): set better defaults for inferencing (#1822)

* fix(defaults): set better defaults for inferencing

This changeset aim to have better defaults and to properly detect when
no inference settings are provided with the model.

If not specified, we defaults to mirostat sampling, and offload all the
GPU layers (if a GPU is detected).

Related to https://github.com/mudler/LocalAI/issues/1373 and https://github.com/mudler/LocalAI/issues/1723

* Adapt tests

* Also pre-initialize default seed
This commit is contained in:
Ettore Di Giacinto 2024-03-13 10:05:30 +01:00 committed by GitHub
parent bc8f648a91
commit f895d06605
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 235 additions and 133 deletions

View file

@ -12,28 +12,23 @@ type PredictionOptions struct {
N int `json:"n"`
// Common options between all the API calls, part of the OpenAI spec
TopP float64 `json:"top_p" yaml:"top_p"`
TopK int `json:"top_k" yaml:"top_k"`
Temperature float64 `json:"temperature" yaml:"temperature"`
Maxtokens int `json:"max_tokens" yaml:"max_tokens"`
Echo bool `json:"echo"`
TopP *float64 `json:"top_p" yaml:"top_p"`
TopK *int `json:"top_k" yaml:"top_k"`
Temperature *float64 `json:"temperature" yaml:"temperature"`
Maxtokens *int `json:"max_tokens" yaml:"max_tokens"`
Echo bool `json:"echo"`
// Custom parameters - not present in the OpenAI API
Batch int `json:"batch" yaml:"batch"`
F16 bool `json:"f16" yaml:"f16"`
IgnoreEOS bool `json:"ignore_eos" yaml:"ignore_eos"`
RepeatPenalty float64 `json:"repeat_penalty" yaml:"repeat_penalty"`
Keep int `json:"n_keep" yaml:"n_keep"`
MirostatETA float64 `json:"mirostat_eta" yaml:"mirostat_eta"`
MirostatTAU float64 `json:"mirostat_tau" yaml:"mirostat_tau"`
Mirostat int `json:"mirostat" yaml:"mirostat"`
FrequencyPenalty float64 `json:"frequency_penalty" yaml:"frequency_penalty"`
TFZ float64 `json:"tfz" yaml:"tfz"`
TypicalP float64 `json:"typical_p" yaml:"typical_p"`
Seed int `json:"seed" yaml:"seed"`
Seed *int `json:"seed" yaml:"seed"`
NegativePrompt string `json:"negative_prompt" yaml:"negative_prompt"`
RopeFreqBase float32 `json:"rope_freq_base" yaml:"rope_freq_base"`