Aller au contenu
ModelPrésentéExterne
Télécharger

Gemma 4 E4B Sol Traces v4

Targeted routing-repair continuation from v3. Continuation-trained on 125 mined trajectories (SWE-bench, synthetic, bash history, GitHub commits) to fix runcommand, searchcode, and applypatch tool selection.

TypeLoRAModèle de baseunsloth/gemma-4-E4B-itVersionadapter_modelBClasse

Source du modèle

Type
LoRA
Taille du fichier
41,6 MB
Modèle de base
unsloth/gemma-4-E4B-it
Version
vadapter_model
Mis à jour
14 oct. 2026
Téléchargements
24
Source
Hugging Face

Description de la source

Gemma-4-E4B-Sol-Traces-v4

Targeted routing-repair continuation from v3. Continuation-trained on 125 mined trajectories (SWE-bench, synthetic, bash history, GitHub commits) to fix run_command, search_code, and apply_patch tool selection.

Training

ParameterValue
Base modelunsloth/gemma-4-E4B-it
Adapter startv3 from-scratch (336 LoRA keys)
Training typeContinuation (100 steps)
Dataset125 train / 15 val / 17 test
SourcesSWE-bench (50), synthetic (50), bash history (55), GitHub (2)
Tool schema6 tools (search_code, run_command, apply_patch, read_file, list_files, write_file)
Learning rate3e-5, cosine, 3% warmup
Batch1 × 8 grad accum
Runtime12 min on H100
Training loss0.093
Eval loss0.671
Cost~$0.80

Routing Improvement

Toolv1v2v3v4
run_command2/50/50/55/5 ✅
search_code0/50/50/51/5
apply_patch1/50/50/50/5
list_files5/55/55/54/5
read_file4/55/55/55/5
no-tool4/55/55/53/5
Overall53%50%50%60%

Files

FileSize
gemma-4-e4b-sol-traces-v4-Q4_K_M.gguf4.97 GiB
gemma-4-e4b-sol-traces-v4-f16.gguf14.02 GiB
adapter/adapter_model.safetensors35 MiB
training_stats.json—

Usage

llama-cli -m gemma-4-e4b-sol-traces-v4-Q4_K_M.gguf -ngl 99

Sources

1 source
Hugging FacePrincipalVersion 2a7ecd769e1609ece1a1cf703a9af7a33222f9c0
huggingface.co/RedTeamLab/Gemma-4-E4B-Sol-Traces-v4/tree/2a7ecd769e1609ece1a1cf703a9af7a33222f9c0

Vérifié 23 août

Artefacts du modèle

1 artefact

adapter/adapter_model.safetensors

safetensors · 41,6 MB · SHA-256 fd6151eefe02…590a · Hugging Face

Download file

Extraits de sources

2 extraits

--- license: gemma language: - en tags: - agentic - tool-calling - function-calling - code-agent - gemma-4 - e4b - lora - unsloth - sol-traces - hermes-agent base_model: unsloth/gemma-4-E4B-it library_name: gguf inference: false --- # Gemma-4-E4B-Sol-Traces-v4 **Targeted routing-repair continuation** from v3. Continuation-trained on 125 mined trajectories (SWE-bench, synthetic, bash history, GitHub commits) to fix `run_command`, `search_code`, and `apply_patch` tool selection. ## Training | Parameter | Value | |---|---| | Base model | `unsloth/gemma-4-E4B-it` | | Adapter start | v3 from-scratch (336 LoRA keys) | | Training type | Continuation (100 steps) | | Dataset | 125 train / 15 val / 17 test | | Sources | SWE-bench (50), synthetic (50), bash history (55), GitHub (2) | | Tool schema | 6 tools (search_code, run_command, apply_patch, read_file, list_files, write_file) | | Learning rate | 3e-5, cosine, 3% warmup | | Batch | 1 × 8 grad accum | | Runtime | 12 min on H100 | | Training loss | 0.093 | | Eval loss | 0.671 | | Cost | ~$0.80 | ## Routing Improvement | Tool | v1 | v2 | v3 | **v4** | |---|---|---|---|---| | run_command | 2/5 | 0/5 | 0/5 | **5/5** ✅ | | search_code | 0/5 | 0/5 | 0/5 | **1/5** | | apply_patch | 1/5 | 0/5 | 0/5 | 0/5 | | list_files | 5/5 | 5/5 | 5/5 | 4/5 | | read_file | 4/5 | 5/5 | 5/5 | 5/5 | | no-tool | 4/5 | 5/5 | 5/5 | 3/5 | | **Overall** | 53% | 50% | 50% | **60%** | ## Files | File | Size | |---|---| | `gemma-4-e4b-sol-traces-v4-Q4_K_M.gguf` | 4.97 GiB | | `gemma-4-e4b-sol-traces-v4-f16.gguf` | 14.02 GiB | | `adapter/adapter_model.safetensors` | 35 MiB | | `training_stats.json` | — | ## Usage ```bash llama-cli -m gemma-4-e4b-sol-traces-v4-Q4_K_M.gguf -ngl 99 ```

Source context: 24 downloads · 0 likes · Library gguf · Repo RedTeamLab/Gemma-4-E4B-Sol-Traces-v4