Skip to content

Commit 28e5f76

Browse files
author
michal.zyga
committed
add test for fine tuning request
1 parent 2b84a1c commit 28e5f76

File tree

3 files changed

+296
-175
lines changed

3 files changed

+296
-175
lines changed

‎core/src/main/scala/sttp/openai/requests/finetuning/FineTuningRequestBody.scala

+183-175
Original file line numberDiff line numberDiff line change
@@ -2,202 +2,210 @@ package sttp.openai.requests.finetuning
22

33
import sttp.openai.OpenAIExceptions.OpenAIException.DeserializationOpenAIException
44
import sttp.openai.json.SnakePickle
5-
import sttp.openai.requests.finetuning.FineTuningRequestBody.Integration.Integration
6-
import sttp.openai.requests.finetuning.FineTuningRequestBody.Method.Method
75
import ujson.Str
86

7+
/** @param model
8+
* The name of the model to fine-tune. You can select one of the supported models
9+
* [[https://platform.openai.com/docs/guides/fine-tuning#which-models-can-be-fine-tuned]].
10+
* @param trainingFile
11+
* The ID of an uploaded file that contains training data. See upload file for how to upload a file. Your dataset must be formatted as a
12+
* JSONL file. Additionally, you must upload your file with the purpose fine-tune. The contents of the file should differ depending on if
13+
* the model uses the chat, completions format, or if the fine-tuning method uses the preference format. See the fine-tuning guide for
14+
* more details.
15+
* @param suffix
16+
* A string of up to 64 characters that will be added to your fine-tuned model name. For example, a suffix of "custom-model-name" would
17+
* produce a model name like ft:gpt-4o-mini:openai:custom-model-name:7p4lURel.
18+
* @param validationFile
19+
* The ID of an uploaded file that contains validation data. If you provide this file, the data is used to generate validation metrics
20+
* periodically during fine-tuning. These metrics can be viewed in the fine-tuning results file. The same data should not be present in
21+
* both train and validation files. Your dataset must be formatted as a JSONL file. You must upload your file with the purpose fine-tune.
22+
* See the fine-tuning guide for more details.
23+
* @param integrations
24+
* A list of integrations to enable for your fine-tuning job.
25+
* @param seed
26+
* The seed controls the reproducibility of the job. Passing in the same seed and job parameters should produce the same results, but may
27+
* differ in rare cases. If a seed is not specified, one will be generated for you.
28+
* @param method
29+
* The method used for fine-tuning.
30+
*/
31+
case class FineTuningRequestBody(
32+
model: FineTuningModel,
33+
trainingFile: String,
34+
suffix: Option[String] = None,
35+
validationFile: Option[String] = None,
36+
integrations: Option[Seq[Integration]] = None,
37+
seed: Option[Int] = None,
38+
method: Option[Method] = None
39+
)
940
object FineTuningRequestBody {
41+
implicit val fineTuningRequestBodyWriter: SnakePickle.Writer[FineTuningRequestBody] = SnakePickle.macroW[FineTuningRequestBody]
42+
}
1043

11-
/** @param model
12-
* The name of the model to fine-tune. You can select one of the supported models
13-
* [[https://platform.openai.com/docs/guides/fine-tuning#which-models-can-be-fine-tuned]].
14-
* @param trainingFile
15-
* The ID of an uploaded file that contains training data. See upload file for how to upload a file. Your dataset must be formatted as
16-
* a JSONL file. Additionally, you must upload your file with the purpose fine-tune. The contents of the file should differ depending
17-
* on if the model uses the chat, completions format, or if the fine-tuning method uses the preference format. See the fine-tuning
18-
* guide for more details.
19-
* @param suffix
20-
* A string of up to 64 characters that will be added to your fine-tuned model name. For example, a suffix of "custom-model-name" would
21-
* produce a model name like ft:gpt-4o-mini:openai:custom-model-name:7p4lURel.
22-
* @param validationFile
23-
* The ID of an uploaded file that contains validation data. If you provide this file, the data is used to generate validation metrics
24-
* periodically during fine-tuning. These metrics can be viewed in the fine-tuning results file. The same data should not be present in
25-
* both train and validation files. Your dataset must be formatted as a JSONL file. You must upload your file with the purpose
26-
* fine-tune. See the fine-tuning guide for more details.
27-
* @param integrations
28-
* A list of integrations to enable for your fine-tuning job.
29-
* @param seed
30-
* The seed controls the reproducibility of the job. Passing in the same seed and job parameters should produce the same results, but
31-
* may differ in rare cases. If a seed is not specified, one will be generated for you.
32-
* @param method
33-
* The method used for fine-tuning.
34-
*/
35-
case class FineTuningRequestBody(
36-
model: FineTuningModel,
37-
trainingFile: String,
38-
suffix: Option[String] = None,
39-
validationFile: Option[String] = None,
40-
integrations: Option[Seq[Integration]] = None,
41-
seed: Option[Int] = None,
42-
method: Option[Method] = None
43-
)
44-
45-
sealed abstract class Type(val value: String)
46-
47-
object Type {
48-
def typeRW(byTypeValue: Map[String, Type]): SnakePickle.ReadWriter[Type] = SnakePickle
49-
.readwriter[ujson.Value]
50-
.bimap[Type](
51-
`type` => SnakePickle.writeJs(`type`.value),
52-
jsonValue =>
53-
SnakePickle.read[ujson.Value](jsonValue) match {
54-
case Str(value) =>
55-
byTypeValue.get(value) match {
56-
case Some(t) => t
57-
case None => throw DeserializationOpenAIException(new Exception(s"Could not deserialize: $value"))
58-
}
59-
case e => throw DeserializationOpenAIException(new Exception(s"Could not deserialize: $e"))
60-
}
61-
)
62-
}
63-
64-
object Method {
65-
66-
object MethodType {
67-
case object Supervised extends Type("supervised")
68-
69-
case object Dpo extends Type("dpo")
70-
71-
private val values: Set[Type] = Set(Supervised, Dpo)
72-
73-
private val byTypeValue = values.map(`type` => `type`.value -> `type`).toMap
74-
75-
implicit val typeRW: SnakePickle.ReadWriter[Type] = Type.typeRW(byTypeValue)
76-
}
77-
78-
/** @param batchSize
79-
* Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower
80-
* variance.
81-
* @param learningRateMultiplier
82-
* Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.
83-
* @param nEpochs
84-
* The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.
85-
* @param beta
86-
* The beta value for the DPO method. A higher beta value will increase the weight of the penalty between the policy and reference
87-
* model.
88-
*/
89-
case class Hyperparameters(
90-
batchSize: Option[Int] = None,
91-
learningRateMultiplier: Option[Float] = None,
92-
nEpochs: Option[Int] = None,
93-
beta: Option[Float] = None
44+
sealed abstract class Type(val value: String)
45+
46+
object Type {
47+
implicit def typeRW(implicit byTypeValue: Map[String, Type]): SnakePickle.ReadWriter[Type] = SnakePickle
48+
.readwriter[ujson.Value]
49+
.bimap[Type](
50+
`type` => SnakePickle.writeJs(`type`.value),
51+
jsonValue =>
52+
SnakePickle.read[ujson.Value](jsonValue) match {
53+
case Str(value) =>
54+
byTypeValue.get(value) match {
55+
case Some(t) => t
56+
case None => throw DeserializationOpenAIException(new Exception(s"Could not deserialize: $value"))
57+
}
58+
case e => throw DeserializationOpenAIException(new Exception(s"Could not deserialize: $e"))
59+
}
9460
)
61+
}
9562

96-
/** @param hyperparameters
97-
* The hyperparameters used for the fine-tuning job.
98-
*/
99-
case class Supervised(
100-
hyperparameters: Option[Hyperparameters] = None
101-
)
63+
/** @param batchSize
64+
* Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower
65+
* variance.
66+
* @param learningRateMultiplier
67+
* Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.
68+
* @param nEpochs
69+
* The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.
70+
* @param beta
71+
* The beta value for the DPO method. A higher beta value will increase the weight of the penalty between the policy and reference model.
72+
*/
73+
case class Hyperparameters(
74+
batchSize: Option[Int] = None,
75+
learningRateMultiplier: Option[Float] = None,
76+
nEpochs: Option[Int] = None,
77+
beta: Option[Float] = None
78+
)
79+
80+
object Hyperparameters {
81+
implicit val hyperparametersW: SnakePickle.Writer[Hyperparameters] = SnakePickle.macroW[Hyperparameters]
82+
}
10283

103-
/** @param hyperparameters
104-
* The hyperparameters used for the fine-tuning job.
105-
*/
106-
case class Dpo(
107-
hyperparameters: Option[Hyperparameters] = None
108-
)
84+
/** @param hyperparameters
85+
* The hyperparameters used for the fine-tuning job.
86+
*/
87+
case class Supervised(
88+
hyperparameters: Option[Hyperparameters] = None
89+
)
10990

110-
/** @param `type`
111-
* The type of method. Is either supervised or dpo.
112-
* @param supervised
113-
* Configuration for the supervised fine-tuning method.
114-
* @param dpo
115-
* Configuration for the DPO fine-tuning method.
116-
*/
117-
case class Method(
118-
`type`: Option[Type] = None,
119-
supervised: Option[Supervised] = None,
120-
dpo: Option[Dpo] = None
121-
)
122-
}
123-
124-
object Integration {
125-
126-
object IntegrationType {
127-
case object Wandb extends Type("wandb")
128-
129-
private val values: Set[Type] = Set(Wandb)
130-
131-
private val byTypeValue = values.map(`type` => `type`.value -> `type`).toMap
132-
133-
implicit val typeRW: SnakePickle.ReadWriter[Type] = Type.typeRW(byTypeValue)
134-
}
135-
136-
/** @param project
137-
* The name of the project that the new run will be created under.
138-
* @param name
139-
* A display name to set for the run. If not set, we will use the Job ID as the name.
140-
* @param entity
141-
* The entity to use for the run. This allows you to set the team or username of the WandB user that you would like associated with
142-
* the run. If not set, the default entity for the registered WandB API key is used.
143-
* @param tags
144-
* A list of tags to be attached to the newly created run. These tags are passed through directly to WandB. Some default tags are
145-
* generated by OpenAI: "openai/finetune", "openai/{base-model}", "openai/{ftjob-abcdef}".
146-
*/
147-
case class Wandb(
148-
project: String,
149-
name: Option[String] = None,
150-
entity: Option[String] = None,
151-
tags: Option[Seq[String]]
152-
)
91+
object Supervised {
92+
implicit val supervisedW: SnakePickle.Writer[Supervised] = SnakePickle.macroW[Supervised]
93+
}
15394

154-
/** @param `type`
155-
* The type of integration to enable. Currently, only "wandb" (Weights and Biases) is supported.
156-
* @param wandb
157-
* The settings for your integration with Weights and Biases. This payload specifies the project that metrics will be sent to.
158-
* Optionally, you can set an explicit display name for your run, add tags to your run, and set a default entity (team, username,
159-
* etc) to be associated with your run.
160-
*/
161-
case class Integration(
162-
`type`: Type,
163-
wandb: Wandb
164-
)
165-
}
95+
/** @param hyperparameters
96+
* The hyperparameters used for the fine-tuning job.
97+
*/
98+
case class Dpo(
99+
hyperparameters: Option[Hyperparameters] = None
100+
)
101+
102+
object Dpo {
103+
implicit val dpoW: SnakePickle.Writer[Dpo] = SnakePickle.macroW[Dpo]
104+
}
166105

167-
sealed abstract class FineTuningModel(val value: String)
106+
/** @param `type`
107+
* The type of method. Is either supervised or dpo.
108+
* @param supervised
109+
* Configuration for the supervised fine-tuning method.
110+
* @param dpo
111+
* Configuration for the DPO fine-tuning method.
112+
*/
113+
case class Method(
114+
`type`: Option[Type] = None,
115+
supervised: Option[Supervised] = None,
116+
dpo: Option[Dpo] = None
117+
)
168118

169-
object FineTuningModel {
119+
object Method {
120+
implicit val methodW: SnakePickle.Writer[Method] = SnakePickle.macroW[Method]
170121

171-
implicit val fineTuningModelRW: SnakePickle.ReadWriter[FineTuningModel] = SnakePickle
172-
.readwriter[ujson.Value]
173-
.bimap[FineTuningModel](
174-
model => SnakePickle.writeJs(model.value),
175-
jsonValue =>
176-
SnakePickle.read[ujson.Value](jsonValue) match {
177-
case Str(value) =>
178-
byFineTuningModelValue.getOrElse(value, CustomFineTuningModel(value))
179-
case e => throw DeserializationOpenAIException(new Exception(s"Could not deserialize: $e"))
180-
}
181-
)
122+
case object Supervised extends Type("supervised")
182123

183-
case object GPT4o20240806 extends FineTuningModel("gpt-4o-2024-08-06")
124+
case object Dpo extends Type("dpo")
125+
126+
private val values: Set[Type] = Set(Supervised, Dpo)
127+
128+
implicit val byTypeValue: Map[String, Type] = values.map(`type` => `type`.value -> `type`).toMap
129+
}
130+
131+
/** @param project
132+
* The name of the project that the new run will be created under.
133+
* @param name
134+
* A display name to set for the run. If not set, we will use the Job ID as the name.
135+
* @param entity
136+
* The entity to use for the run. This allows you to set the team or username of the WandB user that you would like associated with the
137+
* run. If not set, the default entity for the registered WandB API key is used.
138+
* @param tags
139+
* A list of tags to be attached to the newly created run. These tags are passed through directly to WandB. Some default tags are
140+
* generated by OpenAI: "openai/finetune", "openai/{base-model}", "openai/{ftjob-abcdef}".
141+
*/
142+
case class Wandb(
143+
project: String,
144+
name: Option[String] = None,
145+
entity: Option[String] = None,
146+
tags: Option[Seq[String]]
147+
)
148+
149+
object Wandb {
150+
implicit val wandbW: SnakePickle.Writer[Wandb] = SnakePickle.macroW[Wandb]
151+
}
152+
153+
/** @param `type`
154+
* The type of integration to enable. Currently, only "wandb" (Weights and Biases) is supported.
155+
* @param wandb
156+
* The settings for your integration with Weights and Biases. This payload specifies the project that metrics will be sent to.
157+
* Optionally, you can set an explicit display name for your run, add tags to your run, and set a default entity (team, username, etc) to
158+
* be associated with your run.
159+
*/
160+
case class Integration(
161+
`type`: Type,
162+
wandb: Wandb
163+
)
164+
165+
object Integration {
166+
implicit val integrationW: SnakePickle.Writer[Integration] = SnakePickle.macroW[Integration]
167+
168+
case object Wandb extends Type("wandb")
169+
170+
private val values: Set[Type] = Set(Wandb)
171+
172+
private val byTypeValue = values.map(`type` => `type`.value -> `type`).toMap
173+
174+
implicit val typeRW: SnakePickle.ReadWriter[Type] = Type.typeRW(byTypeValue)
175+
}
176+
177+
sealed abstract class FineTuningModel(val value: String)
178+
179+
object FineTuningModel {
180+
181+
implicit val fineTuningModelRW: SnakePickle.ReadWriter[FineTuningModel] = SnakePickle
182+
.readwriter[ujson.Value]
183+
.bimap[FineTuningModel](
184+
model => SnakePickle.writeJs(model.value),
185+
jsonValue =>
186+
SnakePickle.read[ujson.Value](jsonValue) match {
187+
case Str(value) =>
188+
byFineTuningModelValue.getOrElse(value, CustomFineTuningModel(value))
189+
case e => throw DeserializationOpenAIException(new Exception(s"Could not deserialize: $e"))
190+
}
191+
)
184192

185-
case object GPT4oMini20240718 extends FineTuningModel("gpt-4o-mini-2024-07-18")
193+
case object GPT4o20240806 extends FineTuningModel("gpt-4o-2024-08-06")
186194

187-
case object GPT40613 extends FineTuningModel("gpt-4-0613")
195+
case object GPT4oMini20240718 extends FineTuningModel("gpt-4o-mini-2024-07-18")
188196

189-
case object GPT35Turbo0125 extends FineTuningModel("gpt-3.5-turbo-0125")
197+
case object GPT40613 extends FineTuningModel("gpt-4-0613")
190198

191-
case object GPT35Turbo1106 extends FineTuningModel("gpt-3.5-turbo-1106")
199+
case object GPT35Turbo0125 extends FineTuningModel("gpt-3.5-turbo-0125")
192200

193-
case object GPT35Turbo0613 extends FineTuningModel("gpt-3.5-turbo-0613")
201+
case object GPT35Turbo1106 extends FineTuningModel("gpt-3.5-turbo-1106")
194202

195-
case class CustomFineTuningModel(customFineTuningModel: String) extends FineTuningModel(customFineTuningModel)
203+
case object GPT35Turbo0613 extends FineTuningModel("gpt-3.5-turbo-0613")
196204

197-
val values: Set[FineTuningModel] = Set(GPT4o20240806, GPT4oMini20240718, GPT40613, GPT35Turbo0125, GPT35Turbo1106, GPT35Turbo0613)
205+
case class CustomFineTuningModel(customFineTuningModel: String) extends FineTuningModel(customFineTuningModel)
198206

199-
private val byFineTuningModelValue = values.map(model => model.value -> model).toMap
207+
val values: Set[FineTuningModel] = Set(GPT4o20240806, GPT4oMini20240718, GPT40613, GPT35Turbo0125, GPT35Turbo1106, GPT35Turbo0613)
200208

201-
}
209+
private val byFineTuningModelValue = values.map(model => model.value -> model).toMap
202210

203211
}

0 commit comments

Comments
 (0)