llama: preserve field order in user-defined JSON schemas (#8002)

Previously we decoded and re-encoded JSON schemas during validation,
which served no purpose since json.RawMessage already validates JSON
syntax. Worse, the re-encoding lost field ordering from the original
schema, which affects inference quality during step-by-step reasoning.

While fixing this ordering issue by using json.RawMessage directly,
testing revealed that schema_to_grammar (from llama.cpp) also fails to
preserve field order during grammar generation. This appears to be the
root cause of inference degradation.

This change prevents us from mangling the user's original schema order,
but we still need to address the ordering issue in schema_to_grammar.
That will be a separate change.

Updates #7978
This commit is contained in:
Blake Mizerany
2024-12-11 14:07:30 -08:00
committed by GitHub
parent 581a4a5553
commit 9039c821a2
5 changed files with 104 additions and 114 deletions

80
llama/grammar_test.go Normal file
View File

@@ -0,0 +1,80 @@
package llama
import (
"bufio"
"bytes"
"strings"
"testing"
)
// https://github.com/ollama/ollama/issues/7978
const issue7978JSONSchema = `{
"type": "object",
"properties": {
"steps": {
"type": "array",
"items": {
"type": "object",
"properties": {
"explanation": { "type": "string" },
"output": { "type": "string" }
},
"required": ["explanation", "output"],
"additionalProperties": false
}
},
"final_answer": { "type": "string" }
},
"required": ["steps", "final_answer"],
"additionalProperties": false
}`
func TestIssue7978(t *testing.T) {
t.Skip("schema_to_grammar is broken; skipping until fixed")
g := SchemaToGrammar([]byte(issue7978JSONSchema))
if g == nil {
t.Fatal("failed to convert JSON schema to grammar")
}
t.Logf("grammar:\n%s", g)
t.Log()
var sawSteps bool
s := bufio.NewScanner(bytes.NewReader(g))
for s.Scan() {
line := s.Text()
if strings.Contains(line, "steps") {
sawSteps = true
}
if strings.Contains(line, "final-answer") && !sawSteps {
t.Error("expected 'steps' before 'final-answer'")
}
}
}
func TestSchemaToGrammer(t *testing.T) {
t.Skip("schema_to_grammar is broken; skipping until fixed")
cases := []struct {
schema string
prefix []byte // nil is check as nil
}{
{`invalid`, nil},
// Simple heuristic/smoke test
{`{"type":"object"}`, []byte("object ::=")},
}
for _, c := range cases {
t.Run("x", func(t *testing.T) {
g := SchemaToGrammar([]byte(c.schema))
if c.prefix == nil && g != nil {
t.Fatalf("grammar = %v, want nil", g)
}
if !bytes.HasPrefix(g, c.prefix) {
t.Errorf("grammar = %q, want %q", g, c.prefix)
}
})
}
}

View File

@@ -86,12 +86,9 @@ COMPILER inline get_compiler() {
import "C"
import (
"bytes"
_ "embed"
"encoding/json"
"errors"
"fmt"
"log/slog"
"runtime"
"runtime/cgo"
"slices"
@@ -721,21 +718,10 @@ func (s *SamplingContext) Accept(id int, applyGrammar bool) {
C.common_sampler_caccept(s.c, C.llama_token(id), C.bool(applyGrammar))
}
type JsonSchema struct {
Defs map[string]any `json:"$defs,omitempty"`
Properties map[string]any `json:"properties,omitempty"`
Required []string `json:"required,omitempty"`
Title string `json:"title,omitempty"`
Type string `json:"type,omitempty"`
}
func (js JsonSchema) AsGrammar() string {
var b bytes.Buffer
if err := json.NewEncoder(&b).Encode(js); err != nil {
return ""
}
cStr := C.CString(b.String())
// SchemaToGrammar converts the provided JSON schema to a grammar. It returns
// nil if the provided schema is invalid JSON or an invalid JSON schema.
func SchemaToGrammar(schema []byte) []byte {
cStr := C.CString(string(schema))
defer C.free(unsafe.Pointer(cStr))
// Allocate buffer for grammar output with reasonable size
@@ -743,10 +729,10 @@ func (js JsonSchema) AsGrammar() string {
buf := make([]byte, maxLen)
// Call C function to convert schema to grammar
length := C.schema_to_grammar(cStr, (*C.char)(unsafe.Pointer(&buf[0])), C.size_t(maxLen))
if length == 0 {
slog.Warn("unable to convert schema to grammar")
n := C.schema_to_grammar(cStr, (*C.char)(unsafe.Pointer(&buf[0])), C.size_t(maxLen))
if n == 0 {
// preserve nil
return nil
}
return string(buf[:length])
return buf[:n]
}

View File

@@ -1,70 +1 @@
package llama
import (
"strings"
"testing"
"github.com/google/go-cmp/cmp"
)
func TestJsonSchema(t *testing.T) {
testCases := []struct {
name string
schema JsonSchema
expected string
}{
{
name: "empty schema",
schema: JsonSchema{
Type: "object",
},
expected: `array ::= "[" space ( value ("," space value)* )? "]" space
boolean ::= ("true" | "false") space
char ::= [^"\\\x7F\x00-\x1F] | [\\] (["\\bfnrt] | "u" [0-9a-fA-F]{4})
decimal-part ::= [0-9]{1,16}
integral-part ::= [0] | [1-9] [0-9]{0,15}
null ::= "null" space
number ::= ("-"? integral-part) ("." decimal-part)? ([eE] [-+]? integral-part)? space
object ::= "{" space ( string ":" space value ("," space string ":" space value)* )? "}" space
root ::= object
space ::= | " " | "\n" [ \t]{0,20}
string ::= "\"" char* "\"" space
value ::= object | array | string | number | boolean | null`,
},
{
name: "invalid schema with circular reference",
schema: JsonSchema{
Type: "object",
Properties: map[string]any{
"self": map[string]any{
"$ref": "#", // Self reference
},
},
},
expected: "", // Should return empty string for invalid schema
},
{
name: "schema with invalid type",
schema: JsonSchema{
Type: "invalid_type", // Invalid type
Properties: map[string]any{
"foo": map[string]any{
"type": "string",
},
},
},
expected: "", // Should return empty string for invalid schema
},
}
for _, tc := range testCases {
t.Run(tc.name, func(t *testing.T) {
result := tc.schema.AsGrammar()
if !strings.EqualFold(strings.TrimSpace(result), strings.TrimSpace(tc.expected)) {
if diff := cmp.Diff(tc.expected, result); diff != "" {
t.Fatalf("grammar mismatch (-want +got):\n%s", diff)
}
}
})
}
}