Blame for README.md

---
license: apache-2.0
language:
- en
base_model: javierbits/Javi0.3-2B
base_model_relation: finetune
pipeline_tag: image-text-to-text
library_name: transformers
tags:
- qwen3.5
- tool-use
- agent
- weight-space
- dgx-spark
- javi
---

![Javi0.4-2B](https://huggingface.co/javierbits/Javi0.4-2B/resolve/main/javi04-cover.png)

# Javi0.4-2B

Javi0.4-2B is **Javi0.3-2B with one embedding row changed**. No fine-tuning, no new data, no gradients. The tied input/output row for token 51187 (`=open_page`) is multiplied by 0.84. Every other weight is byte-identical to Javi0.3-2B.

The edit biases one decision: after a failed lookup, call `web_search` again instead of opening a URL typed from memory. It does not add a new reasoning capability.

## Measured effect

8 held-out eval-v3 canonical `fail` questions, first `web_search` forced to fail, greedy next tool:

| model | next `web_search` | next `open_page` |
|---|---:|---:|
| Javi0.3-2B | 0/8 | 8/8 |
| Javi0.4-2B | 7/8 | 1/8 |

This probe isolates the decision the edit targets. In a two-seed harness canary with real tools and sampling, 0.84 chose `open_page` after a permanently blocked search and completed, same as the base; it used one extra search on a normal lookup. A 0.75 scale flipped 4/4 but looped on 14 searches under a permanent block, so 0.84 is the conservative setting. The edit reaches the canonical `open_page` tool; in opencode/grok, `webfetch`/`web_fetch` after a 403 is unchanged.

### Full eval v3

175 held-out seeds, 3 runs, date pin 2026-10-07, cached search results. Javi0.3-2B answers on the same seeds are the baseline. Judges: DeepSeek-V4.1-Flash-1017 and mimo-v26-dual.

| metric | Javi0.3-2B | Javi0.4-2B |
|---|---:|---:|
| core pass (judge average) | 0.35 [0.30, 0.40] | 0.41 [0.36, 0.45] |
| core pass (deepseek) | 0.48 | 0.58 |
| core pass (mimo) | 0.23 | 0.24 |
| recovery after a tool failure | 0.55 (n=11) | 0.45 (n=10) |
| made-up links | 0.01 | 0.00 |
| loops | 0.04 | 0.03 |

Paired core pass difference: **+0.06 [0.02, 0.10]** (deepseek +0.10 [0.05, 0.16], mimo +0.02 [-0.02, 0.06]), 95% bootstrap over 175 seeds. The deepseek interval clears zero; the mimo interval crosses zero, so the size of the gain is uncertain. Conflict questions improved (0.33 vs 0.07, n=7) and nodate (0.42 vs 0.17, n=6); premise (0.23 vs 0.29, n=13) and recovery after real failures did not.

Five answer episodes (4 seeds) carry no deepseek verdict because the judge omits the `tools` or `style` key; one seed never finished an episode in any of the three runs and the judge skips it, so judge-only rows cover 174 seeds. Core pass stays at 175 seeds with incomplete episodes scored 0. Full scoreboard: `data/v2/eval/javi04-vs-03-2b/score.md` in the source tree.

## Files

Merged bf16 safetensors (one shard, 4.2 GB), tokenizer, processor files and chat template. The vision tower is Qwen3.5-2B's, unchanged; the edit is in the language model only.

## Run it

Javi was trained and evaluated with thinking off. Send `"chat_template_kwargs": {"enable_thinking": false}` and use temperature 0.2, top-p 0.8, top-k 20 so tool calls stay well formed.

**GGUF** for llama.cpp or lmdeck: [Javi0.4-2B-GGUF](https://huggingface.co/javierbits/Javi0.4-2B-GGUF).

**lmdeck**

```bash
lmdeck use javi0.4-2b
```

**Transformers**

```python
from transformers import AutoModelForImageTextToText, AutoProcessor

model_id = "javierbits/Javi0.4-2B"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForImageTextToText.from_pretrained(model_id, dtype="auto", device_map="auto")

tools = [{"type": "function", "function": {
    "name": "web_search", "description": "Search the web.",
    "parameters": {"type": "object", "properties": {"query": {"type": "string"}}, "required": ["query"]}}}]
messages = [{"role": "user", "content": [{"type": "text", "text": "What's the latest stable Rust release?"}]}]

inputs = processor.apply_chat_template(messages, tools=tools, add_generation_prompt=True, enable_thinking=False,
                                       tokenize=True, return_dict=True, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=512, do_sample=True, temperature=0.2, top_p=0.8, top_k=20)
print(processor.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
```

## The edit

```python
embed_tokens[51187] *= 0.84  # =open_page
```

Token 51187 is a single token in the Qwen3.5 tokenizer. The row is tied between input and output embeddings, so the output logit for `=open_page` is reduced.

## Links

- Base: [Javi0.3-2B](https://huggingface.co/javierbits/Javi0.3-2B)
- 4B control: [Javi0.4-4B](https://huggingface.co/javierbits/Javi0.4-4B)
- Apache 2.0, inherited from [Qwen3.5-2B](https://huggingface.co/Qwen/Qwen3.5-2B).