github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 16:40:08 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #13334] ministral template doesn't handle multiple tool calls #8807

New Issue
Closed
opened 2026-04-12 21:35:23 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 21:35:23 -05:00
Owner
Copy Link

Originally created by @rick-github on GitHub (Dec 5, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13334

Originally assigned to: @pdevine on GitHub.

What is the issue?

This line in the template:

{{- if and (le (len (slice $.Messages $index)) 2) $.Tools }}[AVAILABLE_TOOLS]{{ $.Tools }}[/AVAILABLE_TOOLS]

adds tools to the user message, when the user message is last or second to last message in the message list. So for a completion call with

[
  system message (S)
  user message (U1)
  assistant message (A1)
  user message (U2)
]

the tools get attached to U2 and the model will generate a message (A2) with one or more tool calls. In the process of completing the tools calls, the client will update the message list with A2 and the results of the tool call(s) with a tool message.

[ S U1 A1 U2 A2 T1 ]

Since U2 is no longer the second to last or last message, the tools are not attached in subsequent calls. If the model thinks it needs to make another tool call in order to satisfy the instruction in U2, it cannot. As a concrete example:

[
    {
        "role": "user",
        "content": "what is 2.3 ^ 5.6 and 5.6 ^ 2.3? Now add them together."
    },
    {
        "role": "assistant",
        "tool_calls": [
            "function=Function(name='power', arguments={'x': 2.3, 'y': 5.6})",
            "function=Function(name='power', arguments={'x': 5.6, 'y': 2.3})"
        ]
    },
    {
        "role": "tool",
        "content": "106.0903498817264"
    },
    {
        "role": "tool",
        "content": "52.581438372010965"
    },
    {
        "role": "assistant",
        "content": "Here are the calculations:\n\n- \\( 2.3^{5.6} \\approx 106.0903 \\)\n- \\( 5.6^{2.3} \\approx 52.5814 \\)\n\nAdding them together:\n**106.0903 + 52.5814 = 158.6717**\n\nThe final result is **158.6717**."
    }
]

In contrast, if the user message has tools attached:

[
    {
        "role": "user",
        "content": "what is 2.3 ^ 5.6 and 5.6 ^ 2.3? Now add them together."
    },
    {
        "role": "assistant",
        "tool_calls": [
            "function=Function(name='power', arguments={'x': 2.3, 'y': 5.6})",
            "function=Function(name='power', arguments={'x': 5.6, 'y': 2.3})"
        ]
    },
    {
        "role": "tool",
        "content": "106.0903498817264"
    },
    {
        "role": "tool",
        "content": "52.581438372010965"
    },
    {
        "role": "assistant",
        "tool_calls": [
            "function=Function(name='add', arguments={'l': [106.0903498817264, 52.581438372010965]})"
        ]
    },
    {
        "role": "tool",
        "content": "158.67178825373736"
    },
    {
        "role": "assistant",
        "content": "The result of \\(2.3^{5.6}\\) is approximately **106.09**.\n\nThe result of \\(5.6^{2.3}\\) is approximately **52.58**.\n\nTheir sum is **158.67**."
    }
]

Suggested change to enable ministral to handle multiple tool calls:

--- Modelfile.orig	2025-12-05 02:11:39.391039872 +0100
+++ Modelfile	2025-12-05 02:11:32.070341544 +0100
@@ -2,10 +2,14 @@
 # To build a new Modelfile based on this, replace FROM with:
 FROM ministral-3:14b
 
-TEMPLATE """{{- range $index, $_ := .Messages }}
+TEMPLATE """{{- $lastUserIndex := -1 }}
+{{- range $index, $_ := .Messages }}
+{{- if eq .Role "user" }}{{ $lastUserIndex = $index }}{{ end }}
+{{- end }}
+{{- range $index, $_ := .Messages }}
 {{- if eq .Role "system" }}[SYSTEM_PROMPT]{{ .Content }}[/SYSTEM_PROMPT]
 {{- else if eq .Role "user" }}
-{{- if and (le (len (slice $.Messages $index)) 2) $.Tools }}[AVAILABLE_TOOLS]{{ $.Tools }}[/AVAILABLE_TOOLS]
+{{- if and (eq $lastUserIndex $index) $.Tools }}[AVAILABLE_TOOLS]{{ $.Tools }}[/AVAILABLE_TOOLS]
 {{- end }}[INST]{{ .Content }}[/INST]
 {{- else if eq .Role "assistant" }}
 {{- if .Content }}{{ .Content }}

Relevant log output

OS

Linux, Docker

GPU

Nvidia

CPU

Intel

Ollama version

0.13.1

Originally created by @rick-github on GitHub (Dec 5, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13334 Originally assigned to: @pdevine on GitHub. ### What is the issue? This line in the template: ``` {{- if and (le (len (slice $.Messages $index)) 2) $.Tools }}[AVAILABLE_TOOLS]{{ $.Tools }}[/AVAILABLE_TOOLS] ``` adds tools to the user message, when the user message is last or second to last message in the message list. So for a completion call with ``` [ system message (S) user message (U1) assistant message (A1) user message (U2) ] ``` the tools get attached to U2 and the model will generate a message (A2) with one or more tool calls. In the process of completing the tools calls, the client will update the message list with A2 and the results of the tool call(s) with a `tool` message. ``` [ S U1 A1 U2 A2 T1 ] ``` Since U2 is no longer the second to last or last message, the tools are not attached in subsequent calls. If the model thinks it needs to make another tool call in order to satisfy the instruction in U2, it cannot. As a concrete example: ```json [ { "role": "user", "content": "what is 2.3 ^ 5.6 and 5.6 ^ 2.3? Now add them together." }, { "role": "assistant", "tool_calls": [ "function=Function(name='power', arguments={'x': 2.3, 'y': 5.6})", "function=Function(name='power', arguments={'x': 5.6, 'y': 2.3})" ] }, { "role": "tool", "content": "106.0903498817264" }, { "role": "tool", "content": "52.581438372010965" }, { "role": "assistant", "content": "Here are the calculations:\n\n- \\( 2.3^{5.6} \\approx 106.0903 \\)\n- \\( 5.6^{2.3} \\approx 52.5814 \\)\n\nAdding them together:\n**106.0903 + 52.5814 = 158.6717**\n\nThe final result is **158.6717**." } ] ``` In contrast, if the user message has tools attached: ```json [ { "role": "user", "content": "what is 2.3 ^ 5.6 and 5.6 ^ 2.3? Now add them together." }, { "role": "assistant", "tool_calls": [ "function=Function(name='power', arguments={'x': 2.3, 'y': 5.6})", "function=Function(name='power', arguments={'x': 5.6, 'y': 2.3})" ] }, { "role": "tool", "content": "106.0903498817264" }, { "role": "tool", "content": "52.581438372010965" }, { "role": "assistant", "tool_calls": [ "function=Function(name='add', arguments={'l': [106.0903498817264, 52.581438372010965]})" ] }, { "role": "tool", "content": "158.67178825373736" }, { "role": "assistant", "content": "The result of \\(2.3^{5.6}\\) is approximately **106.09**.\n\nThe result of \\(5.6^{2.3}\\) is approximately **52.58**.\n\nTheir sum is **158.67**." } ] ``` Suggested change to enable ministral to handle multiple tool calls: ```diff --- Modelfile.orig 2025-12-05 02:11:39.391039872 +0100 +++ Modelfile 2025-12-05 02:11:32.070341544 +0100 @@ -2,10 +2,14 @@ # To build a new Modelfile based on this, replace FROM with: FROM ministral-3:14b -TEMPLATE """{{- range $index, $_ := .Messages }} +TEMPLATE """{{- $lastUserIndex := -1 }} +{{- range $index, $_ := .Messages }} +{{- if eq .Role "user" }}{{ $lastUserIndex = $index }}{{ end }} +{{- end }} +{{- range $index, $_ := .Messages }} {{- if eq .Role "system" }}[SYSTEM_PROMPT]{{ .Content }}[/SYSTEM_PROMPT] {{- else if eq .Role "user" }} -{{- if and (le (len (slice $.Messages $index)) 2) $.Tools }}[AVAILABLE_TOOLS]{{ $.Tools }}[/AVAILABLE_TOOLS] +{{- if and (eq $lastUserIndex $index) $.Tools }}[AVAILABLE_TOOLS]{{ $.Tools }}[/AVAILABLE_TOOLS] {{- end }}[INST]{{ .Content }}[/INST] {{- else if eq .Role "assistant" }} {{- if .Content }}{{ .Content }} ``` ### Relevant log output ```shell ``` ### OS Linux, Docker ### GPU Nvidia ### CPU Intel ### Ollama version 0.13.1
GiteaMirror added the toolsbug labels 2026-04-12 21:35:23 -05:00
GiteaMirror commented 2026-04-12 21:35:24 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 5, 2025):

magistral has the same issue.

$ ./ollama-run.py magistral What is 2.3 ^ 5.6 and 5.6 ^ 2.3?  Now multiply them together. --tool power,multiply --system "You MUST use the provided tools to answer the following questions."
[ToolCall(function=Function(name='power', arguments={'x': 2.3, 'y': 5.6}))]
[ToolCall(function=Function(name='power', arguments={'x': 5.6, 'y': 2.3}))]

The values of 2.3^5.6 and 5.6^2.3 are approximately 106.090 and 52.581 respectively. When multiplied together, these numbers result in a product which is 5547.282.

$ ./ollama-run.py magistral:fix What is 2.3 ^ 5.6 and 5.6 ^ 2.3?  Now multiply them together. --tool power,multiply --system "You MUST use the provided tools to answer the following questions."
Let's break down the calculation into steps:

First, I'll calculate \(2.3^{5.6}\) using a calculator or a computational tool.

Then, I'll calculate \(5.6^{2.3}\) in the same manner.

Finally, I'll multiply these two results together to get the desired outcome.

Let's proceed with the calculations:
[ToolCall(function=Function(name='power', arguments={'x': 2.3, 'y': 5.6}))]

Now I'll calculate 5.6 to the power of 2.3.
[ToolCall(function=Function(name='power', arguments={'x': 5.6, 'y': 2.3}))]

Now that I have both results, I'll multiply them together: 106.0903498817264 and 52.581438372010965.
[ToolCall(function=Function(name='multiply', arguments={'x': 106.0903498817264, 'y': 52.581438372010965}))]

The final result of multiplying \(2.3^{5.6}\) and \(5.6^{2.3}\) is approximately 5578.3832.
<!-- gh-comment-id:3615046706 --> @rick-github commented on GitHub (Dec 5, 2025): [magistral](https://ollama.com/library/magistral) has the same issue. ```console $ ./ollama-run.py magistral What is 2.3 ^ 5.6 and 5.6 ^ 2.3? Now multiply them together. --tool power,multiply --system "You MUST use the provided tools to answer the following questions." [ToolCall(function=Function(name='power', arguments={'x': 2.3, 'y': 5.6}))] [ToolCall(function=Function(name='power', arguments={'x': 5.6, 'y': 2.3}))] The values of 2.3^5.6 and 5.6^2.3 are approximately 106.090 and 52.581 respectively. When multiplied together, these numbers result in a product which is 5547.282. ``` ```console $ ./ollama-run.py magistral:fix What is 2.3 ^ 5.6 and 5.6 ^ 2.3? Now multiply them together. --tool power,multiply --system "You MUST use the provided tools to answer the following questions." Let's break down the calculation into steps: First, I'll calculate \(2.3^{5.6}\) using a calculator or a computational tool. Then, I'll calculate \(5.6^{2.3}\) in the same manner. Finally, I'll multiply these two results together to get the desired outcome. Let's proceed with the calculations: [ToolCall(function=Function(name='power', arguments={'x': 2.3, 'y': 5.6}))] Now I'll calculate 5.6 to the power of 2.3. [ToolCall(function=Function(name='power', arguments={'x': 5.6, 'y': 2.3}))] Now that I have both results, I'll multiply them together: 106.0903498817264 and 52.581438372010965. [ToolCall(function=Function(name='multiply', arguments={'x': 106.0903498817264, 'y': 52.581438372010965}))] The final result of multiplying \(2.3^{5.6}\) and \(5.6^{2.3}\) is approximately 5578.3832. ```
GiteaMirror commented 2026-04-12 21:35:25 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 10, 2025):

This should be fixed now. Thanks @rick-github !

<!-- gh-comment-id:3635126183 --> @pdevine commented on GitHub (Dec 10, 2025): This should be fixed now. Thanks @rick-github !
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug tools
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#8807
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?