Views
No views yet
tokenizer.chat_template) is extended to render the tool_calls field of
assistant messages, and the files marked in the table additionally embed
LiquidAI's recommended sampling as general.sampling.* metadata.pip install nobodywho1from nobodywho import Chat
2
3chat = Chat("huggingface:NobodyWho/LFM2.5-1.2B-Instruct-GGUF/LFM2.5-1.2B-Instruct-Q8_0-vendor-sampling.gguf")
4response = chat.ask("What is the capital of Denmark?").completed()
5print(response) # Copenhagen!1from nobodywho import Chat, tool
2
3@tool(description="Gets the current weather for a city")
4def get_weather(city: str) -> str:
5 return f"It is sunny and 22°C in {city}."
6
7chat = Chat(
8 "huggingface:NobodyWho/LFM2.5-1.2B-Instruct-GGUF/LFM2.5-1.2B-Instruct-Q8_0-vendor-sampling.gguf",
9 tools=[get_weather],
10)
11print(chat.ask("What is the weather in Paris?").completed())[!NOTE] Tool calling with LFM models ships in the upcomingnobodywhorelease (PR #564). These files also work in any other llama.cpp-based runtime; the original unmodified GGUFs live in the upstream LiquidAI/LFM2.5-1.2B-Instruct-GGUF repo.
message.content. Runtimes that store tool
calls in the structured tool_calls field (the HF "unified tool use"
convention, used by NobodyWho and OpenAI-style APIs) re-render assistant
tool-call turns as empty turns, so the model never sees its own previous
calls — causing re-issued tool calls and degraded multi-turn tool use.<|tool_call_start|>[get_weather(city="Paris")]<|tool_call_end|>| File | Fix recipe | NobodyWho tool-suite score |
|---|---|---|
LFM2.5-1.2B-Instruct-Q8_0-vendor-sampling.gguf | template + vendor sampling | 14/14 |
LFM2.5-1.2B-Instruct-F16.gguf | template only | 14/14 |
LFM2.5-1.2B-Instruct-Q4_0-vendor-sampling.gguf | template + vendor sampling | 12/14 (double-calls two tests) |
general.sampling.* metadata, taken from the vendor's LEAP deployment config:
temperature 0.3, min_p 0.15, repetition_penalty 1.05. The F16
deliberately ships without sampling metadata: at full precision, embedding
those values drops the bash-writing test (13/14 vs 14/14 with default
sampling), so runtimes fall back to their own defaults.The embedded values were previouslytemp 0.1, top_k 50(the vendor's model-card prose, which conflicts with its LEAP config) and have been corrected to the LEAP values above. All scores re-verified against these files with the corrected sampler: Q8_0 14/14, F16 14/14 (sampling-free), F16 + vendor sampling 13/14, Q4_0 12/14 (double-calls two tests).
| Property | Value |
|---|---|
| Parameters | 1.2B |
| Context length | 128,000 tokens |
| License | LFM Open License v1.0 |
| Base model | LiquidAI/LFM2.5-1.2B-Instruct |