Use the full model name on the Hub

README.md

@@ -14,9 +14,9 @@ tags:
   - exaone
 ---
 
-# EXAONE Finance
+# EXAONE Forecast for Finance
 
-**EXAONE Finance** is a time series foundation model (TSFM) built for financial
+**EXAONE Forecast for Finance** is a time series foundation model (TSFM) built for financial
 forecasting. It replaces self-attention with two linear-time operators — a causal
 1D convolution for temporal mixing and a group-aware pooling MLP for variate
 mixing — so cost grows linearly in both sequence length and variate count. A
@@ -34,7 +34,7 @@ and is released under a different license — see [License](#license).
 
 ## Overview
 
-How EXAONE Finance differs from other time series foundation models, in the two
+How EXAONE Forecast for Finance differs from other time series foundation models, in the two
 choices that set a TSFM's cost and its inductive bias — how it mixes information
 along time, and along variates — and in what it was pretrained on.
 
@@ -44,9 +44,9 @@ along time, and along variates — and in what it was pretrained on.
 | Chronos, Chronos-Bolt, TimesFM, Lag-Llama, TimeGPT, Timer, MOMENT, Sundial, TEMPO, ROSE, Time-MoE, PatchTST-FM, Kairos, YingLong, CleanTS, TabPFN-TS, VisionTS | Attention | — | ❌ |
 | TiRex (xLSTM), TempoPFN, FlowState, Reverso | RNN / SSM | — | ❌ |
 | TTM (TinyTimeMixer) | MLP | MLP | ❌ |
-| **EXAONE Finance (Ours)** | **CNN** | **MLP** | ✅ |
+| **EXAONE Forecast for Finance (Ours)** | **CNN** | **MLP** | ✅ |
 
-EXAONE Finance is the only **attention-free convolutional** TSFM in this
+EXAONE Forecast for Finance is the only **attention-free convolutional** TSFM in this
 comparison, and the only one pretrained on **financial-domain data**, which we
 generate synthetically. Both of its mixers are linear-time, so cost grows
 linearly in sequence length and in variate count rather than quadratically.
@@ -82,7 +82,7 @@ placements (lower is better). 44 models were scored on identical windows.
 
 | # | Model | Tier 1 (Point) | Avg. Rank | Tier 2 (IC) | Avg. Rank | Tier 3 (Portfolio) | Avg. Rank | Rank Sum |
 |---|---|---|---|---|---|---|---|---|
-| **1** | **EXAONE Finance** | **1** | **6.53** | **1** | **7.45** | **1** | **7.43** | **3** |
+| **1** | **EXAONE Forecast for Finance** | **1** | **6.53** | **1** | **7.45** | **1** | **7.43** | **3** |
 | 2 | Chronos-2 (Synthetic) | 9 | 11.68 | 3 | 8.74 | 2 | 9.90 | 14 |
 | 3 | Reverso (Small) | 6 | 9.60 | 2 | 8.64 | 6 | 13.04 | 14 |
 | 4 | TiRex-1.1 | 2 | 6.98 | 4 | 9.71 | 14 | 15.76 | 20 |
@@ -91,7 +91,7 @@ placements (lower is better). 44 models were scored on identical windows.
 The three tiers score complementary objectives: **Tier 1** point accuracy (MASE,
 hit rate), **Tier 2** cross-sectional ranking skill (information coefficient), and
 **Tier 3** realized portfolio performance (return, Sharpe, volatility, maximum
-drawdown). EXAONE Finance ranks first in all three for a perfect rank sum of 3,
+drawdown). EXAONE Forecast for Finance ranks first in all three for a perfect rank sum of 3,
 ahead of the strongest baseline at 14, and is the only model that wins its
 head-to-head comparison against all 43 baselines (per-opponent win rate
 0.51–0.90). At 202M parameters it sits on the Pareto frontier, outperforming
@@ -180,9 +180,8 @@ Third-party open source components and their licenses are listed in
 }
 ```
 
-> The technical report introduces this model as **EXAONE Forecast for Finance**
-> in its abstract. **EXAONE Finance** is the name used everywhere else, in the
-> report, on the Hub, and in the code.
+> The code package and its API use the short form **EXAONE Finance**
+> (`from exaone_forecast.finance import from_pretrained`).
 
 ## Contact