Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-19 09:15:18 +02:00
Code Issues Packages Projects Releases Wiki Activity

an easy way to toggle thinking for jinja

Browse Source
This commit is contained in:
Concedo
2026-05-21 22:45:33 +08:00
parent e8bf5b9c6c
commit 2451feaf69
1 changed files with 53 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
koboldcpp.py
+53 -6
View File
@@ -2534,7 +2534,7 @@ def sd_comfyui_tranform_params(genparams):
return genparams
# json with top-level dict
def parse_json_object(value, field):
def parse_json_object(value, field, silent=False):
if not value:
return None
broken = False
@@ -2569,14 +2569,20 @@ def parse_json_object(value, field):
if value:
try:
import ast
value = ast.literal_eval(value)
# Convert JS literals to Python literals
fixed = re.sub(r'\btrue\b', 'True', value)
fixed = re.sub(r'\bfalse\b', 'False', fixed)
fixed = re.sub(r'\bnull\b', 'None', fixed)
value = ast.literal_eval(fixed)
if value and isinstance(value, dict):
return value
except Exception:
pass
print(f"Warning: couldn't parse {field} field.")
if not silent:
print(f"Warning: couldn't parse {field} field.")
else:
print(f"Warning: {field} field - not a JSON object.")
if not silent:
print(f"Warning: {field} field - not a JSON object.")
return None
def gendefaults_parse_meta_field(value):
@@ -7597,6 +7603,7 @@ def show_gui():
jinja_var = ctk.IntVar(value=0)
jinja_tools_var = ctk.IntVar(value=0)
jinja_kwargs_var = ctk.StringVar()
jinja_think_var = ctk.StringVar(value="default")
moeexperts_var = ctk.StringVar(value=str(-1))
moecpu_var = ctk.StringVar(value=str(0))
defaultgenamt_var = ctk.StringVar(value=str(default_genlen))
@@ -8355,16 +8362,46 @@ def show_gui():
jinjatoolsbox.grid()
jinjakwargsbox.grid()
jinjakwargsboxlbl.grid()
jinjathinkbox.grid()
jinjathinklbl.grid()
else:
jinja_tools_var.set(0)
jinjatoolsbox.grid_remove()
jinjakwargsbox.grid_remove()
jinjakwargsboxlbl.grid_remove()
jinjathinkbox.grid_remove()
jinjathinklbl.grid_remove()
changed_gpulayers_estimate()
def togglejinjathink(a,b,c):
curr = parse_json_object(jinja_kwargs_var.get(),"tempjinja")
curr = (curr if curr else {})
changed = False
if jinja_think_var.get()=="true":
curr["enable_thinking"] = True
changed = True
elif jinja_think_var.get()=="false":
curr["enable_thinking"] = False
changed = True
elif "enable_thinking" in curr:
del curr["enable_thinking"]
changed = True
if changed:
jinja_kwargs_var.set(json.dumps(curr) if curr else "")
def updatejinjathinktoggle(a,b,c):
curr = parse_json_object(jinja_kwargs_var.get(),"tempjinja",True)
curr = (curr if curr else {})
if "enable_thinking" in curr:
if curr["enable_thinking"] is True:
jinja_think_var.set("true")
elif curr["enable_thinking"] is False:
jinja_think_var.set("false")
makecheckbox(context_tab, "Use Jinja", jinja_var, row=45, command=togglejinja, tooltiptxt="Enables using jinja chat template formatting for chat completions endpoint. Other endpoints are unaffected.")
jinjatoolsbox = makecheckbox(context_tab, "Jinja for Tools", jinja_tools_var, row=45 ,padx=(140), tooltiptxt="Allows jinja even with tool calls. If unchecked, jinja will be disabled when tools are used.")
jinjakwargsbox,jinjakwargsboxlbl = makelabelentry(context_tab, "Jj.Kwargs:", jinja_kwargs_var, row=45, width=80, padx=(350), singleline=True, tooltip='Set additiona fields for Jinja JSON template parser, must be a valid json object.\nSpecified as JSON fields: {"KEY1":"VALUE1", "KEY2":"VALUE2"...}', labelpadx=285)
jinja_think_choices = ['default', 'true', 'false']
jinjathinkbox, jinjathinklbl = makelabelcombobox(context_tab, "Jinja Thinking:", jinja_think_var, 45, command=togglejinjathink,labelpadx=(280), padx=370, width=100, tooltiptxt="Tries to enable or disable thinking in Jinja mode. This is a shortcut to setting Jinja Kwargs directly.", values=jinja_think_choices)
jinjakwargsbox,jinjakwargsboxlbl = makelabelentry(context_tab, "Jinja Kwargs:", jinja_kwargs_var, row=47, width=200, padx=(100), singleline=True, tooltip='Set additiona fields for Jinja JSON template parser, must be a valid json object.\nSpecified as JSON fields: {"KEY1":"VALUE1", "KEY2":"VALUE2"...}')
jinja_var.trace_add("write", togglejinja)
jinja_kwargs_var.trace_add("write", updatejinjathinktoggle)
makelabelentry(context_tab, "MoE Experts:", moeexperts_var, row=55, padx=(86), singleline=True, tooltip="Override number of MoE experts.")
moecpu_box,moecpu_box_lbl = makelabelentry(context_tab, "MoE CPU Layers:", moecpu_var, row=55, padx=(334), singleline=True, tooltip="Force Mixture of Experts (MoE) weights of the first N layers to the CPU.\nSetting it higher than GPU layers has no effect.", labelpadx=(230))
makelabelentry(context_tab, "Override KV:", override_kv_var, row=57, padx=(86), singleline=True, width=130, tooltip="Override metadata value by key. Separate multiple values with commas. Format is name=type:value. Types: int, float, bool, str")
@@ -8649,6 +8686,7 @@ def show_gui():
togglesdlora(1,1,1)
togglejinja(1,1,1)
toggleadmin(1,1,1)
updatejinjathinktoggle(1,1,1)
# launch
def guilaunch():
@@ -9038,10 +9076,18 @@ def show_gui():
nobostoken_var.set(mydict["nobostoken"] if ("nobostoken" in mydict) else 0)
jinja_var.set(mydict["jinja"] if ("jinja" in mydict) else 0)
jinja_tools_var.set(mydict["jinja_tools"] if ("jinja_tools" in mydict) else 0)
jinja_think_var.set("default")
jinja_kwargs = (mydict["jinja_kwargs"] if ("jinja_kwargs" in mydict and mydict["jinja_kwargs"]) else "")
if isinstance(jinja_kwargs, type({})):
jinja_kwargs = json.dumps(jinja_kwargs)
jinja_kwargs_var.set(jinja_kwargs)
if "jinjathink" in mydict:
jinja_kwargs = (json.loads(jinja_kwargs) if jinja_kwargs else {})
if mydict["jinjathink"]=="true":
jinja_kwargs["enable_thinking"] = True
if mydict["jinjathink"]=="false":
jinja_kwargs["enable_thinking"] = False
jinja_kwargs = json.dumps(jinja_kwargs) if jinja_kwargs else ""
jinja_kwargs_var.set(jinja_kwargs if jinja_kwargs else "")
jinjatemplate_var.set(mydict["jinjatemplate"] if ("jinjatemplate" in mydict and mydict["jinjatemplate"]) else "")
enableguidance_var.set(mydict["enableguidance"] if ("enableguidance" in mydict) else 0)
@@ -11599,6 +11645,7 @@ if __name__ == '__main__':
advparser.add_argument("--jinja_tools","--jinja-tools","--jinjatools", help="Enables using jinja chat template formatting for chat completions endpoint. Other endpoints are unaffected. Tool calls are done with jinja.", action='store_true')
advparser.add_argument("--jinja_kwargs","--jinja-kwargs","--jinjakwargs","--chat-template-kwargs", metavar=('{"parameter":"value",...}'), help="Set additional fields for Jinja JSON template parser, must be a valid JSON object.", default="")
advparser.add_argument("--jinjatemplate","--chat-template-file", metavar=('[filename]'), help="Select a custom Jinja chat template, will overwrite model jinja chat template", default="")
advparser.add_argument("--jinjathink", help="A quick way to enable or disable thinking in the jinja template.", type=str, choices=['default','true','false'], default="default")
advparser.add_argument("--noflashattention","--no-flash-attn","-nofa", help="Disables flash attention.", action='store_true')
advparser.add_argument("--lowvram","-nkvo","--no-kv-offload", help="If supported by the backend, do not offload KV to GPU (lowvram mode). Not recommended, will be slow.", action='store_true')
advparser.add_argument("--quantkv", help="Sets the KV cache data type quantization, options are f16/bf16/q8_0/q5_1/q4_0. Requires Flash Attention for full effect, otherwise only K cache is quantized.",metavar=('[quantization level f16/bf16/q8_0/q5_1/q4_0]'), type=str, choices=["f16","bf16","q8_0","q5_1","q4_0","0","1","2","3"], default="f16")