File size: 15,272 Bytes
4927bea
 
 
 
 
fc98b5b
 
4927bea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fc98b5b
 
 
 
 
 
 
 
 
4927bea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fc98b5b
4927bea
 
 
 
 
 
 
 
 
fc98b5b
 
4927bea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
809ac78
4927bea
 
 
 
 
 
 
 
809ac78
 
 
 
4927bea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
809ac78
4927bea
 
 
 
809ac78
4927bea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
809ac78
4927bea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
809ac78
 
4927bea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
"""
server.py β€” MicroJulia OpenAI-compatible inference server
Serves POST /v1/chat/completions (streaming + non-streaming) and GET /v1/models.

Loads the pure-Julia MicroGPT char-level model from best_model.json on HF Hub.
Architecture: Pre-norm GPT with RMSNorm (no learnable params), ReLU MLP,
separate Q/K/V attention.  5K params, 28-char vocab, val_loss=2.34.

Follows the RandyGPT FastAPI/uvicorn pattern for proven HF Spaces compatibility.
"""

import json
import math
import time
import uuid
import os
import torch
import torch.nn as nn
import torch.nn.functional as F
from pathlib import Path
from fastapi import FastAPI, HTTPException, Request
from fastapi.responses import JSONResponse, StreamingResponse
from fastapi.middleware.cors import CORSMiddleware
from fastapi.exceptions import RequestValidationError
from pydantic import BaseModel
from typing import List, Optional
from huggingface_hub import hf_hub_download


# ── Model definition ──────────────────────────────────────────────────────────
# Matches the Julia AutoGrad training code exactly:
#   Pre-norm blocks with RMSNorm (no learnable weights), ReLU MLP,
#   separate Q/K/V attention, no final norm before lm_head.


def rms_norm(x, eps=1e-5):
    """RMSNorm without learnable parameters (matches Julia rmsnorm_ag)."""
    return x * torch.rsqrt(x.pow(2).mean(dim=-1, keepdim=True) + eps)


class Attn(nn.Module):
    def __init__(self, n_embd, n_head):
        super().__init__()
        self.n_head = n_head
        self.head_dim = n_embd // n_head
        self.scale = 1.0 / math.sqrt(self.head_dim)
        self.wq = nn.Linear(n_embd, n_embd, bias=False)
        self.wk = nn.Linear(n_embd, n_embd, bias=False)
        self.wv = nn.Linear(n_embd, n_embd, bias=False)
        self.wo = nn.Linear(n_embd, n_embd, bias=False)

    def forward(self, x):
        B, T, C = x.shape
        q = self.wq(x).view(B, T, self.n_head, self.head_dim).transpose(1, 2)
        k = self.wk(x).view(B, T, self.n_head, self.head_dim).transpose(1, 2)
        v = self.wv(x).view(B, T, self.n_head, self.head_dim).transpose(1, 2)
        scores = q @ k.transpose(-2, -1) * self.scale
        mask = torch.full((T, T), float('-inf'), device=x.device).triu(1)
        attn = F.softmax(scores + mask, dim=-1)
        out = (attn @ v).transpose(1, 2).contiguous().view(B, T, C)
        return self.wo(out)


class MLP(nn.Module):
    def __init__(self, n_embd):
        super().__init__()
        self.fc1 = nn.Linear(n_embd, 4 * n_embd, bias=False)
        self.fc2 = nn.Linear(4 * n_embd, n_embd, bias=False)

    def forward(self, x):
        return self.fc2(F.relu(self.fc1(x)))


class Block(nn.Module):
    def __init__(self, n_embd, n_head):
        super().__init__()
        self.attn = Attn(n_embd, n_head)
        self.mlp = MLP(n_embd)

    def forward(self, x):
        x = x + self.attn(rms_norm(x))
        x = x + self.mlp(rms_norm(x))
        return x


class MicroGPT(nn.Module):
    def __init__(self, vocab_size, n_embd, n_head, n_layer, block_size):
        super().__init__()
        self.block_size = block_size
        self.wte = nn.Embedding(vocab_size, n_embd)
        self.wpe = nn.Embedding(block_size, n_embd)
        self.layers = nn.ModuleList([Block(n_embd, n_head) for _ in range(n_layer)])
        self.lm_head = nn.Linear(n_embd, vocab_size, bias=False)

    def forward(self, ids):
        B, T = ids.shape
        x = self.wte(ids) + self.wpe(torch.arange(T, device=ids.device).unsqueeze(0))
        for block in self.layers:
            x = block(x)
        return self.lm_head(x)

    @torch.no_grad()
    def generate_stream(self, ids, max_new_tokens=200, temperature=0.1,
                        top_k=8, repetition_penalty=1.3, valid_vocab=None):
        """Yields (token_id, is_last) one token at a time."""
        self.eval()
        generated = []
        for i in range(max_new_tokens):
            ctx = ids[:, -self.block_size:]
            logits = self(ctx)[:, -1, :]  # (1, vocab_size)
            logits = logits[0]  # (vocab_size,)

            # Mask out any token indices beyond the actual charset
            if valid_vocab is not None and logits.shape[0] > valid_vocab:
                logits[valid_vocab:] = float('-inf')

            # Repetition penalty
            if repetition_penalty > 1.0:
                seen = set()
                for t in generated[-self.block_size:]:
                    seen.add(t)
                for t in ctx[0].tolist():
                    seen.add(t)
                for t in seen:
                    if 0 <= t < logits.shape[0]:
                        if logits[t] > 0:
                            logits[t] /= repetition_penalty
                        else:
                            logits[t] *= repetition_penalty

            # Temperature
            logits = logits / max(temperature, 0.01)

            # Top-k filtering
            if top_k > 0 and top_k < logits.shape[0]:
                topk_vals, _ = torch.topk(logits, top_k)
                logits[logits < topk_vals[-1]] = float('-inf')

            probs = F.softmax(logits, dim=-1)
            nxt = torch.multinomial(probs, 1)
            ids = torch.cat([ids, nxt.view(1, 1)], dim=1)
            token_id = nxt.item()
            generated.append(token_id)
            is_last = (i == max_new_tokens - 1)
            yield token_id, is_last

    @torch.no_grad()
    def generate(self, ids, max_new_tokens=200, temperature=0.1,
                 top_k=8, repetition_penalty=1.3, valid_vocab=None):
        """Generate all tokens at once, return full id sequence."""
        self.eval()
        generated = []
        for token_id, _ in self.generate_stream(ids, max_new_tokens, temperature,
                                                 top_k, repetition_penalty, valid_vocab):
            generated.append(token_id)
        return generated


# ── Char-level tokenizer ──────────────────────────────────────────────────────

class CharTokenizer:
    def __init__(self, uchars):
        self.uchars = uchars
        self.stoi = {c: i for i, c in enumerate(uchars)}
        self.itos = {i: c for i, c in enumerate(uchars)}
        self.vocab_size = len(uchars)

    def encode(self, text):
        """Encode text to token IDs (char-level, lowercase, skip unknown)."""
        return [self.stoi[c] for c in text.lower() if c in self.stoi]

    def decode(self, ids):
        """Decode token IDs back to text."""
        return "".join(self.itos.get(i, "?") for i in ids)


# ── Load model at startup ────────────────────────────────────────────────────

REPO = os.environ.get("HF_REPO", "LisaMegaWatts/JuliaGPT")
MODEL_ID = "microjulia-philosophy"
DEVICE = "cpu"

print(f"Loading MicroJulia model from {REPO} ...")
ckpt_path = hf_hub_download(repo_id=REPO, filename="best_model.json")

with open(ckpt_path) as f:
    ckpt = json.load(f)

hp = ckpt["hyperparams"]
n_embd = hp["n_embd"]
n_head = hp["n_head"]
n_layer = hp["n_layer"]
block_size = hp["block_size"]
sd = ckpt["state_dict"]

# Determine vocab_size from weight dimensions
wte_weights = torch.tensor(sd["wte"], dtype=torch.float32)
vocab_size = wte_weights.shape[0]

# Build char tokenizer from embedded uchars
tok = CharTokenizer(ckpt["uchars"])

print(f"  n_embd={n_embd}, n_head={n_head}, n_layer={n_layer}, block_size={block_size}")
print(f"  vocab_size={vocab_size} (weights), chars={tok.vocab_size} ({tok.uchars})")

if "training" in ckpt:
    t = ckpt["training"]
    print(f"  trained: {t.get('total_steps_completed', '?')} steps, "
          f"best_val_loss={t.get('best_val_loss', '?'):.4f}")

# Build model and load weights
model = MicroGPT(vocab_size, n_embd, n_head, n_layer, block_size)

state = {}
state["wte.weight"] = wte_weights
state["wpe.weight"] = torch.tensor(sd["wpe"], dtype=torch.float32)
state["lm_head.weight"] = torch.tensor(sd["lm_head"], dtype=torch.float32)

for i in range(n_layer):
    prefix = f"layer{i}"
    state[f"layers.{i}.attn.wq.weight"] = torch.tensor(sd[f"{prefix}.attn_wq"], dtype=torch.float32)
    state[f"layers.{i}.attn.wk.weight"] = torch.tensor(sd[f"{prefix}.attn_wk"], dtype=torch.float32)
    state[f"layers.{i}.attn.wv.weight"] = torch.tensor(sd[f"{prefix}.attn_wv"], dtype=torch.float32)
    state[f"layers.{i}.attn.wo.weight"] = torch.tensor(sd[f"{prefix}.attn_wo"], dtype=torch.float32)
    state[f"layers.{i}.mlp.fc1.weight"] = torch.tensor(sd[f"{prefix}.mlp_fc1"], dtype=torch.float32)
    state[f"layers.{i}.mlp.fc2.weight"] = torch.tensor(sd[f"{prefix}.mlp_fc2"], dtype=torch.float32)

model.load_state_dict(state)
model.eval()
print(f"Model ready β€” {sum(p.numel() for p in model.parameters())} params")


# ── FastAPI app ───────────────────────────────────────────────────────────────

app = FastAPI(title="MicroJulia", version="1.0.0")

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"],
)


def _openai_error(status: int, message: str, err_type: str = "invalid_request_error", code: str = None):
    body = {"error": {"message": message, "type": err_type}}
    if code:
        body["error"]["code"] = code
    return JSONResponse(status_code=status, content=body)


@app.exception_handler(HTTPException)
async def http_exception_handler(request: Request, exc: HTTPException):
    return _openai_error(exc.status_code, str(exc.detail))


@app.exception_handler(RequestValidationError)
async def validation_exception_handler(request: Request, exc: RequestValidationError):
    msg = "; ".join(f"{e['loc'][-1]}: {e['msg']}" for e in exc.errors())
    return _openai_error(422, msg, code="invalid_request_error")


@app.get("/")
def root():
    return {
        "name": "MicroJulia",
        "version": "1.0.0",
        "description": "Pure Julia char-level GPT trained on classical philosophy",
        "architecture": "MicroGPT (no LayerNorm, GELU, separate Q/K/V)",
        "model": {
            "vocab_size": tok.vocab_size,
            "n_embd": n_embd,
            "n_layer": n_layer,
            "n_head": n_head,
            "block_size": block_size,
            "params": sum(p.numel() for p in model.parameters()),
        },
        "endpoints": ["/v1/models", "/v1/chat/completions"],
        "features": ["streaming", "OpenAI-compatible"],
    }


@app.get("/v1/models")
def list_models():
    return {
        "object": "list",
        "data": [{
            "id": MODEL_ID,
            "object": "model",
            "created": 1700000000,
            "owned_by": "microjulia",
        }]
    }


class Message(BaseModel):
    role: str
    content: str

class ChatRequest(BaseModel):
    model: Optional[str] = MODEL_ID
    messages: List[Message]
    max_tokens: Optional[int] = 200
    temperature: Optional[float] = 0.1
    top_k: Optional[int] = 8
    repetition_penalty: Optional[float] = 1.3
    n: Optional[int] = 1
    stream: Optional[bool] = False


def _sse(data: dict) -> str:
    return f"data: {json.dumps(data)}\n\n"


def _stream_completion(ids, max_tokens, temperature, top_k, rep_penalty,
                       completion_id, _model, _tok):
    """Generator that yields SSE chunks one token at a time."""
    token_count = 0

    # Initial chunk with role
    yield _sse({
        "id": completion_id,
        "object": "chat.completion.chunk",
        "created": int(time.time()),
        "model": MODEL_ID,
        "choices": [{
            "index": 0,
            "delta": {"role": "assistant", "content": ""},
            "finish_reason": None,
        }],
    })

    for token_id, is_last in _model.generate_stream(
        ids, max_new_tokens=max_tokens,
        temperature=temperature, top_k=top_k,
        repetition_penalty=rep_penalty, valid_vocab=_tok.vocab_size
    ):
        token_text = _tok.decode([token_id])
        token_count += 1
        finish_reason = ("length" if token_count >= max_tokens else "stop") if is_last else None

        yield _sse({
            "id": completion_id,
            "object": "chat.completion.chunk",
            "created": int(time.time()),
            "model": MODEL_ID,
            "choices": [{
                "index": 0,
                "delta": {"content": token_text},
                "finish_reason": finish_reason,
            }],
        })

    yield "data: [DONE]\n\n"


@app.post("/v1/chat/completions")
def chat_completions(req: ChatRequest):
    _m, _t = model, tok

    prompt = req.messages[-1].content.strip() if req.messages else ""
    if not prompt:
        raise HTTPException(status_code=400, detail="No content in messages")

    ids = _t.encode(prompt)
    if not ids:
        # If prompt has no valid chars, start with a random token
        ids = [0]

    max_tokens = max(1, min(req.max_tokens or 200, block_size))
    temperature = max(0.01, min(req.temperature or 0.1, 2.0))
    top_k = max(1, min(req.top_k or 8, tok.vocab_size))
    rep_penalty = max(1.0, min(req.repetition_penalty or 1.3, 3.0))
    n = max(1, min(req.n or 1, 4))
    completion_id = f"chatcmpl-{uuid.uuid4().hex[:8]}"

    tensor = torch.tensor([ids], dtype=torch.long)

    # ── Streaming ─────────────────────────────────────────────────────────
    if req.stream:
        return StreamingResponse(
            _stream_completion(tensor, max_tokens, temperature, top_k,
                               rep_penalty, completion_id, _m, _t),
            media_type="text/event-stream",
            headers={"X-Accel-Buffering": "no"},
        )

    # ── Non-streaming ─────────────────────────────────────────────────────
    choices = []
    total_completion_tokens = 0

    for i in range(n):
        generated = _m.generate(tensor.clone(), max_new_tokens=max_tokens,
                                temperature=temperature, top_k=top_k,
                                repetition_penalty=rep_penalty,
                                valid_vocab=_t.vocab_size)
        text = _t.decode(generated)
        total_completion_tokens += len(generated)
        choices.append({
            "index": i,
            "message": {"role": "assistant", "content": text},
            "finish_reason": "length" if len(generated) >= max_tokens else "stop",
        })

    return {
        "id": completion_id,
        "object": "chat.completion",
        "created": int(time.time()),
        "model": MODEL_ID,
        "system_fingerprint": "microjulia-v1",
        "choices": choices,
        "usage": {
            "prompt_tokens": len(ids),
            "completion_tokens": total_completion_tokens,
            "total_tokens": len(ids) + total_completion_tokens,
        },
    }