Downloads · 30 days
6
14% of all-time downloads
MANGSEOK123/qwen3-4b-tau2-telecom-v2-6-1ep
qwen3-4b-tau2-telecom-v2-6-1ep is a machine learning model from MANGSEOK123. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Qwen3-4B-Instruct-2507 consolidated with OEL on the tau2-bench telecom domain, using 46 externally authored task–memory pairs (recall 20 / composition 17 / substitution 9).
Downloads · 30 days
6
14% of all-time downloads
All-time downloads
44
Public
Parameters
4.4B
8.8 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors8.8 GB · 100%
From the Hugging Face model README
Qwen3-4B-Instruct-2507 consolidated with OEL on the tau2-bench telecom domain, using 46 externally authored task–memory pairs (recall 20 / composition 17 / substitution 9).
| base model | Qwen3-4B-Instruct-2507 |
| training pairs | 46 |
| batch size | 12, every batch full |
| epochs | 1 |
| learning rate | 3e-6, constant |
| gradient clipping | 1.0 (verl default) |
| loss | full KL over all response tokens, kl_topk 256 |
| user simulator | gpt-4.1-mini (temperature 0) |
The student replays each task without memory; the teacher is the same weights with that task's memory in the system prompt. Only the prompt differs, and no reward is used.
| step | KL loss | entropy | grad norm |
|---|---|---|---|
| 1 | 0.021 | 0.209 | 7.064 |
| 2 | 0.040 | 0.262 | 6.215 |
| 3 | 0.011 | 0.349 | 3.671 |
| 4 | 0.014 | 0.191 | 6.642 |
Each step reads a different batch, so the loss column tracks batch difficulty rather than convergence.
Not evaluated. Pushed straight after training.
For context on the signal these memories carry: on a 59-pair telecom training set the
base model scores avg 0.085 with no memory and 0.175 with each task's memory in the
prompt — successes double, about 3.1 SE over 177 episodes. On the tau2-bench telecom
test split the base model scores avg 0.056 / pass@4 0.175.
vllm serve MANGSEOK123/qwen3-4b-tau2-telecom-v2-6-1ep \
--enable-auto-tool-choice --tool-call-parser hermes --max-model-len 40960