ModelListedExternal

Fabliq 8B Agent FromBase Reasoning

> Reasoning-expanded sibling of Fabliq-8B-Agent-FromBase. Adds general + deep reasoning (WithinUs + Helio) on top of the FromBase agentic foundation. Mirrors Fabliq-8B-Agent-Reasoning but for the FromBase ablation...

Sourcehuggingface.co/LLM-OS-Models/Fabliq-8B-Agent-FromBase-Reasoning/tree/dd84a4d5050487178a9743444250880988843e67

KindText modelBase modelLLM-OS-Models/Fabliq-8B-Agent-FromBaseVersionmodel

Model source

Kind: Text model
File size: 15.8 GB
Base model: LLM-OS-Models/Fabliq-8B-Agent-FromBase
Version: vmodel
Updated: Oct 16, 2026
Source: Hugging Face

Source description

Fabliq-8B-Agent-FromBase-Reasoning 🌊🔬🧠

Reasoning-expanded sibling of Fabliq-8B-Agent-FromBase. Adds general + deep reasoning (WithinUs + Helio) on top of the FromBase agentic foundation. Mirrors Fabliq-8B-Agent-Reasoning but for the FromBase ablation lineage.

🔬 Why this variant?

Two parallel lineages, both end with reasoning expansion:

Lineage	Phase-1	Phase-2
ToolBench line	`Fabliq-8B-Agent` (ToolBench → Fable-5)	`Fabliq-8B-Agent-Reasoning`
FromBase line (this)	`Fabliq-8B-Agent-FromBase` (base → Fable-5)	`Fabliq-8B-Agent-FromBase-Reasoning` (this)

Comparing the two Phase-2 endpoints answers: does ToolBench foundation matter for the final reasoning model, or is Fable-5 + reasoning enough?

🧪 Model details


Architecture	Lfm2MoeForCausalLM (24 layers, 32 experts, 4 experts/token)
Parameters	~8B total / ~1B active (MoE)
Context	8,192 trained · 128K native
Precision	bfloat16
Fine-tune type	Full-parameter SFT continuation
License	Apache 2.0

📚 Training data (Phase-2B only)

Same as Phase-2: WithinUs 135 + Helio 146 = 281 rows × 4 epoch.

🔧 Training procedure

Hyperparameter	Value
Base	`LLM-OS-Models/Fabliq-8B-Agent-FromBase`
Schedule	4 epochs, LR 3e-7
Max sequence length	8,192
GPUs	8× H200 (effective batch 64)
Train runtime	~6 min

🌳 Model tree

LiquidAI/LFM2.5-8B-A1B
  ├─ ToolBench line:
  │   └─ LFM2.5-8B-A1B-Terminal-ToolBench-Full-SFT-1Epoch
  │       └─ Fabliq-8B-Agent
  │           └─ Fabliq-8B-Agent-Reasoning
  └─ FromBase line (this):
      └─ Fabliq-8B-Agent-FromBase
          └─ Fabliq-8B-Agent-FromBase-Reasoning  ← this model

📜 License

Apache 2.0.

Sources

1 source

Hugging FacePrimaryVersion dd84a4d5050487178a9743444250880988843e67

huggingface.co/LLM-OS-Models/Fabliq-8B-Agent-FromBase-Reasoning/tree/dd84a4d5050487178a9743444250880988843e67

Verified Oct 16

Model artifacts

1 artifact

model.safetensors

safetensors · 15.8 GB · SHA-256 9bf24b9a23cc…e5b1 · Hugging Face

Download

Source excerpts

2 excerpts

ReadmeOct 16huggingface.co/LLM-OS-Models/Fabliq-8B-Agent-FromBase-Reasoning

--- library_name: transformers license: apache-2.0 base_model: LLM-OS-Models/Fabliq-8B-Agent-FromBase base_model_relation: finetune tags: - lfm - liquid-ai - moe - agentic - terminal - reasoning - fable-5 - withinus - helio - distillation - sft - ablation language: - en pipeline_tag: text-generation --- # Fabliq-8B-Agent-FromBase-Reasoning 🌊🔬🧠 > **Reasoning-expanded sibling** of [Fabliq-8B-Agent-FromBase](https://huggingface.co/LLM-OS-Models/Fabliq-8B-Agent-FromBase). Adds general + deep reasoning (WithinUs + Helio) on top of the FromBase agentic foundation. Mirrors [Fabliq-8B-Agent-Reasoning](https://huggingface.co/LLM-OS-Models/Fabliq-8B-Agent-Reasoning) but for the FromBase ablation lineage. ## 🔬 Why this variant? Two parallel lineages, both end with reasoning expansion: | Lineage | Phase-1 | Phase-2 | | --- | --- | --- | | **ToolBench line** | `Fabliq-8B-Agent` (ToolBench → Fable-5) | `Fabliq-8B-Agent-Reasoning` | | **FromBase line (this)** | `Fabliq-8B-Agent-FromBase` (base → Fable-5) | `Fabliq-8B-Agent-FromBase-Reasoning` (this) | Comparing the two Phase-2 endpoints answers: **does ToolBench foundation matter for the final reasoning model, or is Fable-5 + reasoning enough?** ## 🧪 Model details | | | | --- | --- | | **Architecture** | Lfm2MoeForCausalLM (24 layers, 32 experts, 4 experts/token) | | **Parameters** | ~8B total / ~1B active (MoE) | | **Context** | 8,192 trained · 128K native | | **Precision** | bfloat16 | | **Fine-tune type** | Full-parameter SFT continuation | | **License** | Apache 2.0 | ## 📚 Training data (Phase-2B only) Same as Phase-2: WithinUs 135 + Helio 146 = 281 rows × 4 epoch. ## 🔧 Training procedure | Hyperparameter | Value | | --- | --- | | Base | `LLM-OS-Models/Fabliq-8B-Agent-FromBase` | | Schedule | 4 epochs, LR 3e-7 | | Max sequence length | 8,192 | | GPUs | 8× H200 (effective batch 64) | | Train runtime | ~6 min | ## 🌳 Model tree ``` LiquidAI/LFM2.5-8B-A1B ├─ ToolBench line: │ └─ LFM2.5-8B-A1B-Terminal-ToolBench-Full-SFT-1Epoch │ └─ Fabliq-8B-Agent │ └─ Fabliq-8B-Agent-Reasoning └─ FromBase line (this): └─ Fabliq-8B-Agent-FromBase └─ Fabliq-8B-Agent-FromBase-Reasoning ← this model ``` ## 📜 License Apache 2.0.

NoteOct 16huggingface.co/LLM-OS-Models/Fabliq-8B-Agent-FromBase-Reasoning

Source context: 172 downloads · 12 likes · Pipeline text-generation · Library transformers · Repo LLM-OS-Models/Fabliq-8B-Agent-FromBase-Reasoning