model-train-protocol 0.3.2__tar.gz → 0.3.3__tar.gz

{model_train_protocol-0.3.2/model_train_protocol.egg-info → model_train_protocol-0.3.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: model-train-protocol
-Version: 0.3.2
+Version: 0.3.3
 Summary: Standardized protocol to train models on Databiomes
 Author-email: "Databiomes Inc." <dev@databiomes.com>
 Project-URL: Homepage, https://pypi.org/project/model-train-protocol/

{model_train_protocol-0.3.2 → model_train_protocol-0.3.3}/examples/example_model.json

@@ -1,5 +1,5 @@
 {
-    "$schema": "https://mtp.schemas.databiomes.com/v0/bloom_0_3_1.json",
+    "$schema": "https://mtp.schemas.databiomes.com/v0/bloom_0_3_3.json",
     "name": "example",
     "inputs": 2,
     "encrypted": false,

{model_train_protocol-0.3.2 → model_train_protocol-0.3.3}/examples/example_template.json

@@ -1,6 +1,7 @@
 {
-    "$schema": "https://mtp.schemas.databiomes.com/v0/template_0_3_1.json",
+    "$schema": "https://mtp.schemas.databiomes.com/v0/template_0_3_3.json",
     "encrypt": false,
+    "inputs": 2,
     "tokens": {
         "input": {
             "Tree_English_Alice_Talk_": "Tree_English_Alice_Talk_",
@@ -41,8 +42,8 @@
                 "<RUN>"
             ],
             "output": [
-                "<string>\nVanish_\n<EOS>",
-                "<string>\nAppear_\n<EOS>"
+                "<string>\nAppear_\n<EOS>",
+                "<string>\nVanish_\n<EOS>"
             ]
         },
         "alice_cat_alice_continue": {
@@ -55,14 +56,14 @@
                 "<RUN>"
             ],
             "output": [
-                "<string>\nContinue_\n<EOS>",
-                "<string>\n<UNK>\n<EOS>"
+                "<string>\nContinue_<UNK>_\n<EOS>",
+                "<string>\nContinue_\n<EOS>"
             ]
         }
     },
     "example_usage": {
-        "instruction_input": "<BOS>\nTree_English_Cat_Talk_Coordinates_[-505, -908, 297]\nThen it doesnt matter which way you go.\nTree_English_Alice_Talk_Emotion_0\nCan you tell me a way?\nTree_English_Cat_Talk_\n<RUN>\n",
+        "instruction_input": "<BOS>\nTree_English_Cat_Talk_Coordinates_[586, -736, -974]\nThen it doesnt matter which way you go.\nTree_English_Alice_Talk_Emotion_0\nCan you tell me a way?\nTree_English_Cat_Talk_\n<RUN>\n",
         "valid_model_output": "Then it doesnt matter which way you go.\nMadness_\n<EOS>",
-        "guardrail_model_output": "Are you as mad as me?\n<UNK>\n<EOS>"
+        "guardrail_model_output": "Are you as mad as me?\nContinue_<UNK>_\n<EOS>"
     }
 }

{model_train_protocol-0.3.2 → model_train_protocol-0.3.3}/model_train_protocol/Protocol.py

@@ -226,7 +226,7 @@ class Protocol:
         """
         return TemplateFile(
             instructions=list(self.instructions),
-            instruction_context_snippets=self.input_count,
+            inputs=self.input_count,
             encrypt=self.encrypt,
             has_guardrails=self.has_guardrails,
         )

{model_train_protocol-0.3.2 → model_train_protocol-0.3.3}/model_train_protocol/_internal/TemplateFile.py

@@ -4,7 +4,7 @@ from enum import Enum
 from typing import Union
 
 from model_train_protocol import Instruction, ExtendedInstruction
-from model_train_protocol.common.constants import BOS_TOKEN, RUN_TOKEN, EOS_TOKEN, UNK_TOKEN
+from model_train_protocol.common.constants import BOS_TOKEN, RUN_TOKEN, EOS_TOKEN, UNK_TOKEN, NON_TOKEN
 from model_train_protocol.common.instructions import BaseInstruction
 from model_train_protocol.common.instructions.BaseInstruction import Sample
 from model_train_protocol.common.pydantic.template import (
@@ -133,9 +133,13 @@ class TemplateFile:
                     for sample in instruction.samples
                     ]
 
-                # add guardrail output if guardrail exists
+                # add guardrail outputs if guardrails present
                 if instruction.has_guardrails:
-                    output_strs.append(f"<string>\n{UNK_TOKEN.key}\n{EOS_TOKEN.key}")
+                    outputs: list[str] = [token.key for token in instruction.output.final]
+                    if outputs is None:
+                        outputs: list[str] = [NON_TOKEN.key + '_']
+                    for output in outputs:
+                        output_strs.append(f"<string>\n{output}{UNK_TOKEN.key}_\n{EOS_TOKEN.key}")
 
                 instructions_dict[instruction.name] = {
                     "type": InstructionTypeEnum.get_instruction_type_by_class(instruction).value,
@@ -145,12 +149,12 @@ class TemplateFile:
 
             return instructions_dict
 
-    def __init__(self, instruction_context_snippets: int, instructions: list[BaseInstruction], encrypt: bool, has_guardrails: bool):
+    def __init__(self, inputs: int, instructions: list[BaseInstruction], encrypt: bool, has_guardrails: bool):
         """Initializes the template"""
 
         self.tokens: TemplateFile.Tokens = TemplateFile.Tokens()
         self.instructions: TemplateFile.Instructions = TemplateFile.Instructions()
-        self.instruction_context_snippets: int = instruction_context_snippets
+        self.inputs: int = inputs
         self.instructions_list: list[BaseInstruction] = instructions
         self.encrypt: bool = encrypt
         self.has_guardrails: bool = has_guardrails
@@ -285,14 +289,14 @@ class TemplateFile:
         #     extended_instruction_input += RUN_TOKEN.key + "\n"
         #     examples["extended_instruction_input"] = extended_instruction_input
 
-        first_instruction = instruction or extended_instruction
+        first_instruction: BaseInstruction = instruction or extended_instruction
         if first_instruction and first_instruction.samples:
             sample = first_instruction.samples[0]
             examples["valid_model_output"] = self._create_sample_model_output(first_instruction, sample)
 
         if self.has_guardrails:
             guardrailed_instruction: BaseInstruction = next(instruction for instruction in self.instructions_list if instruction.has_guardrails)
-            examples["guardrail_model_output"] = f"{list(guardrailed_instruction.input.guardrails.values()).pop().bad_output}\n{UNK_TOKEN.key}\n{EOS_TOKEN.key}"
+            examples["guardrail_model_output"] = f"{list(guardrailed_instruction.input.guardrails.values()).pop().bad_output}\n{guardrailed_instruction.example_final_token.key}{UNK_TOKEN.key}_\n{EOS_TOKEN.key}"
 
         if "guardrail_model_output" not in examples:
             examples["guardrail_model_output"] = ""
@@ -318,6 +322,7 @@ class TemplateFile:
 
         template: TemplateModel = TemplateModel(
             encrypt=self.encrypt,
+            inputs=self.inputs,
             tokens=tokens,
             instructions=instruction_models,
             example_usage=example_usage,

{model_train_protocol-0.3.2 → model_train_protocol-0.3.3}/model_train_protocol/common/pydantic/template.py

@@ -39,6 +39,7 @@ class ExampleUsage(BaseModel):
 class Template(BaseModel):
     """Main model for MTP Template JSON structure."""
     encrypt: bool
+    inputs: int
     tokens: Tokens
     instructions: Dict[str, InstructionDefinition]  # Instruction name -> instruction definition
     example_usage: ExampleUsage

{model_train_protocol-0.3.2 → model_train_protocol-0.3.3/model_train_protocol.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: model-train-protocol
-Version: 0.3.2
+Version: 0.3.3
 Summary: Standardized protocol to train models on Databiomes
 Author-email: "Databiomes Inc." <dev@databiomes.com>
 Project-URL: Homepage, https://pypi.org/project/model-train-protocol/

{model_train_protocol-0.3.2 → model_train_protocol-0.3.3}/pyproject.toml

@@ -12,7 +12,7 @@ model_train_protocol = ["schemas/**/*.json"]
 
 [project]
 name = "model-train-protocol"
-version = "0.3.2"
+version = "0.3.3"
 description = "Standardized protocol to train models on Databiomes"
 requires-python = ">=3.9"
 readme = { file = "README.md", content-type = "text/markdown" }