README.md

---
library_name: transformers
pipeline_tag: text-generation
base_model: deepseek-ai/deepseek-coder-1.3b-instruct
tags:
  - coding
  - sql
  - postgresql
  - text-to-sql
  - local
  - schema-grounded
  - interchained
---

# Imagine v8

**Imagine** is a compact local coding model built and fine-tuned by **Interchained**.

It is focused on PostgreSQL, schema-grounded text-to-SQL, database reasoning, and local-first deployment.

Imagine v8 is a research checkpoint based on `deepseek-ai/deepseek-coder-1.3b-instruct`.

## What Imagine is for

Imagine is designed for:

- PostgreSQL generation
- schema-grounded text-to-SQL
- database reasoning
- read-only SQL workflows
- local coding assistance
- structured database tasks
- local and self-hosted inference

Imagine is intentionally more focused than a general-purpose chatbot.

## Identity

The deployed model identity is **Imagine**.

Imagine was built and fine-tuned by **Interchained**.

DeepSeek-Coder is part of the upstream model lineage, but the deployed model identity is Imagine.

## Local-first

Imagine is intended to run locally on hardware controlled by the user.

Normal inference does not require a metered cloud inference API.

A larger teacher model may be used during research or curriculum generation, but Imagine does not require that teacher at runtime.

## PostgreSQL specialization

Imagine is trained around PostgreSQL and schema-grounded query generation.

The project emphasizes whether generated SQL:

1. parses correctly
2. follows the read-only contract
3. binds against the supplied schema
4. executes successfully
5. answers the requested question

The goal is not to match one reference SQL string exactly.

Different SQL queries can be semantically equivalent.

## Read-only SQL contract

Imagine is trained for read-only database tasks.

Expected SQL includes:

```sql
SELECT ...
```

and read-only:

```sql
WITH ...
SELECT ...
```

The database-task contract excludes mutating and destructive statements such as:

- INSERT
- UPDATE
- DELETE
- MERGE
- TRUNCATE
- ALTER
- DROP
- CREATE
- COPY

Production deployments should still enforce read-only access at the database-permission level.

## Schema grounding

Imagine should use only the schema supplied in the prompt.

It should not invent:

- tables
- columns
- relationships
- foreign keys
- join keys
- identifiers

If the requested information cannot be derived from the supplied schema, Imagine should avoid fabricating an answer.

Depending on the request, the correct response may be `UNANSWERABLE` or a clarification request.

## Ambiguity handling

Imagine is trained not to guess through material ambiguity.

If two reasonable interpretations would produce meaningfully different SQL or results, Imagine should ask for clarification rather than silently choosing one.

## Output protocol

SQL tasks may use structured sentinel output:

```text
<<<SQL>>>
SELECT ...
<<<END>>>
```

Unsupported requests may use:

```text
<<<UNANSWERABLE>>>
...
<<<END>>>
```

Ambiguous requests may use clarification behavior rather than guessing.

## Training

Imagine v8 was produced through iterative full fine-tuning.

The training path includes:

- DeepSeek-Coder 1.3B Instruct base
- PostgreSQL-focused supervised fine-tuning
- execution-gated SQL curriculum generation
- identity training
- adversarial identity examples
- schema-grounding examples
- protocol-focused training
- expanded system-contract training

The project uses real PostgreSQL execution as part of its evaluation and curriculum pipeline.

## Execution gate

SQL can be checked through several levels:

### Parse

The query must parse as PostgreSQL.

### Safety

The query must satisfy the read-only contract.

### Bind

The query must resolve successfully against the real supplied schema.

This catches nonexistent tables, nonexistent columns, ambiguous references, and related schema errors.

### Execute

The query must execute successfully against PostgreSQL.

### Result agreement

The result can be compared with a reference query.

Reference-result agreement is useful, but it is not identical to semantic correctness. A different projection or implementation may still answer the user's question correctly.

## Evaluation

A held-out telemetry evaluation contains 21 PostgreSQL questions.

A checkpoint in the v6 lineage achieved:

- 100% strict parse rate
- 100% execute rate
- 0 invented relations
- 0 invented columns
- 0 refusals
- 15/21 strict reference-result agreements
- 71.4% strict execution-agreement score

Some reference mismatches involved valid executable SQL that differed from the reference query in projection or implementation.

For that reason, Imagine research distinguishes between:

- protocol compliance
- PostgreSQL executability
- exact result agreement
- semantic correctness

## Example

Prompt:

```text
What is the average cpu_pct in samples?
```

Possible output:

```text
<<<SQL>>>
SELECT AVG(cpu_pct) FROM samples;
<<<END>>>
```

## Loading Imagine

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "Interchained/imagine-v8"

tokenizer = AutoTokenizer.from_pretrained(model_id)

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

messages = [
    {
        "role": "user",
        "content": "What model am I speaking with?"
    }
]

inputs = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_tensors="pt",
).to(model.device)

with torch.no_grad():
    output = model.generate(
        inputs,
        max_new_tokens=128,
        do_sample=False,
    )

response = tokenizer.decode(
    output[0][inputs["input_ids"].shape[-1]:],
    skip_special_tokens=True,
)

print(response)
```

## Intended use

Imagine is intended for:

- developers
- database engineers
- AI engineers
- PostgreSQL users
- text-to-SQL researchers
- local-model experimentation

## Limitations

Imagine v8 is a research checkpoint.

It may still:

- generate incorrect SQL
- misunderstand ambiguous requests
- apply incorrect filters
- choose incorrect ordering
- misunderstand business semantics
- deviate from the intended output protocol
- produce incomplete answers
- show residual behavior from the upstream base model

Generated SQL should be reviewed before use in important systems.

## Security

Do not rely on model behavior alone for database safety.

Production systems should enforce controls such as:

- read-only database roles
- statement timeouts
- row limits
- query validation
- schema restrictions
- application-level authorization
- audit logging

## Project philosophy

Imagine is built around three ideas:

**Grounding over guessing.**  
The supplied schema is the source of truth.

**Execution over string similarity.**  
Correctness should be judged by what SQL actually does, not only whether it matches one reference string.

**Local control over unnecessary dependency.**  
Useful coding models should be able to run on infrastructure controlled by the user.

## Status

**Imagine v8 — Research Checkpoint**

Built and fine-tuned by **Interchained**.